System

Abstract

A system is provided.SOLUTION: A specialized education platform for adapting a general artificial intelligence to a specific specialized field, comprising: means for uploading training data selected by an industry expert; means for checking consistency of the uploaded training data and storing the uploaded training data; means for setting a training target and a standard based on the specialized field; means for generating and initializing a learning model of the general artificial intelligence based on the set training target and standard; means for performing an iterative learning session on the learning model using the training data; and means for evaluating performance of the artificial intelligence after the learning session, the system includes a means for notifying the user of the result, a means for adjusting the learning model based on feedback from the user, and a means for recognizing the artificial intelligence as a professional when the training is completed and the target is achieved.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] General artificial intelligence (AI) can be applied to many purposes, but to perform tasks with high accuracy in specific specialized fields, specialized training is required. However, conventional methods make it difficult to adapt AI to specific fields, and the training process is inefficient. In addition, there are few ways to grasp the results of AI training in real time, making it impossible to quickly incorporate feedback from experts. This has made appropriate training and management for utilizing AI in specialized jobs a challenge. [Means for solving the problem]

[0005] In order to solve the above problems, the present invention provides a specialized education platform for adapting a general-purpose artificial intelligence to a specific specialized field. The present invention includes the following components.

[0006] 1. A means to upload training data selected by industry experts;

[0007] 2. A means of verifying and storing the integrity of uploaded training data;

[0008] 3. Means of establishing discipline-based training goals and standards;

[0009] 4. A means for generating and initializing a learning model of the general artificial intelligence based on established training goals and criteria;

[0010] 5. Means for performing repeated training sessions on the learning model using the training data;

[0011] 6. A means of evaluating the performance of the AI ​​after a training session and informing the user of the results;

[0012] 7. A means to adjust the learning model based on user feedback;

[0013] 8. A means of professional certification of artificial intelligence once training is complete and goals are met.

[0014] This allows general AI to be specialized in specific fields, enabling efficient training processes and real-time performance evaluation, and allowing for rapid feedback from users to improve the quality of specialized AI.

[0015] "General purpose artificial intelligence" is artificial intelligence that can perform a wide range of tasks that are not limited to a specific application or field.

[0016] A "specialized education platform" is a system for providing education and training to adapt general artificial intelligence to specific specialized fields.

[0017] "Training data" is a dataset used to train artificial intelligence, and is material containing information and knowledge related to a particular field of expertise.

[0018] An "industry expert" is an individual or organization with advanced knowledge and experience in a particular area of ​​expertise.

[0019] "Uploading means" is a function that allows a user to send data from their own terminal to the server.

[0020] "Integrity" refers to a state in which data is consistent, accurate, and consistent.

[0021] "Training goals" are specific standards of performance or ability that artificial intelligence must achieve through training.

[0022] "Standards" refer to rules, indicators, and standards used as reference in training and evaluation.

[0023] A "learning model" is a framework of knowledge and algorithms that a general AI builds internally based on training data to carry out a specific task.

[0024] "Initialization" is the process of setting parameters and other parameters to their initial settings in order to start training a learning model.

[0025] An "iterative learning session" is a series of processes in which an AI repeatedly uses training data to learn.

[0026] "Means for evaluating performance" are methods or functions that measure how effectively a trained AI can perform tasks against set goals.

[0027] "Means for notifying" is a function for informing users of the results of training and evaluation.

[0028] "Means for adjusting based on feedback" refers to methods or functions that modify and optimize the learning model or training process based on user evaluations and opinions.

[0029] "Certification" is the process of officially recognizing that an AI has met its set training goals. [Brief explanation of the drawings]

[0030] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram illustrating a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0031] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0032] First, the terms used in the following description will be explained.

[0033] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0034] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0035] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0036] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0037] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0038] [First embodiment]

[0039] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0040] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0041] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0042] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0043] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0044] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0045] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0046] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0047] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0048] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0049] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0050] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0051] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0052] Overall system configuration

[0053] Server: The core of the system, it receives, stores, processes data, generates, trains, and evaluates learning models. The server manages multiple AI learning models and serves as a specialized education platform.

[0054] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc.

[0055] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[0056] Program processing

[0057] 1. Data upload and management

[0058] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0059] Terminal: The user uploads the selected training data file to the server via the terminal.

[0060] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[0061] 2. Setting training goals and standards

[0062] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0063] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[0064] Server: Receives training goals and criteria and stores them in a database.

[0065] 3. Generating and initializing the learning model

[0066] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the generated learning model and performs initial settings (e.g., setting initial parameter values, setting hyperparameters for training).

[0067] 4. Running and monitoring the training process

[0068] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[0069] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[0070] User: View the AI's performance during training on a dashboard and provide feedback as needed. If the training data or criteria need to be adjusted, send new instructions to the server.

[0071] 5. Evaluation and feedback

[0072] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[0073] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0074] 6. Certification and deployment of professional AI

[0075] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0076] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0077] Specific examples

[0078] For example, if a publisher develops a medical translation AI, the following embodiment applies:

[0079] 1. Data upload:

[0080] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[0081] 2. Setting training goals and standards:

[0082] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[0083] 3. Generating and initializing the learning model:

[0084] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[0085] 4. Running the training process:

[0086] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0087] 5. Evaluation and feedback:

[0088] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[0089] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[0090] 6. Certification and deployment of professional AI:

[0091] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[0092] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0093] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[0094] The processing flow will be explained below.

[0095] Step 1:

[0096] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0097] Step 2:

[0098] Device: The user uploads the selected training data file to the server via the device. An upload UI is provided to assist with file selection and transmission.

[0099] Step 3:

[0100] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the data and automatically inspects the format and content of the data before storing it.

[0101] Step 4:

[0102] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0103] Step 5:

[0104] Terminal: Provides a UI for inputting training goals and criteria, sends the settings to the server, and displays a confirmation message to the user when the sending operation is complete.

[0105] Step 6:

[0106] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[0107] Step 7:

[0108] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[0109] Step 8:

[0110] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[0111] Step 9:

[0112] Server: Automatically evaluates the AI's performance after each training session, records the results, scores them based on the evaluation criteria, and stores them in a database.

[0113] Step 10:

[0114] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[0115] Step 11:

[0116] User: Check the AI's performance during training on the dashboard and provide feedback as needed. Feedback is sent from the device to the server.

[0117] Step 12:

[0118] Server: After each training session, the server evaluates the AI's performance and uses the feedback to retune the learning model, resetting parameters as needed.

[0119] Step 13:

[0120] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0121] Step 14:

[0122] Server: Receives new settings from the user, adjusts the training plan again, and continues the learning session.

[0123] Step 15:

[0124] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0125] Step 16:

[0126] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​works effectively in the field.

[0127] Example 1

[0128] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0129] Adapting general-purpose AI to specific specialized fields requires effective management of diverse training data and the generation of highly accurate learning models. However, conventional systems have not adequately verified the consistency of specialized training data, monitored the training process in real time, or applied the data to a practical environment after the training goal has been achieved. As a result, it has been difficult to efficiently and effectively develop specialized AI and deploy it in practice.

[0130] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0131] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model of the general purpose AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is completed and the goal is achieved, and means for exporting the generated professional AI model for application to a practical environment. This makes it possible to efficiently and effectively adapt the general purpose AI to a specific field of expertise and to develop and deploy professional AI that is suited to practical use.

[0132] An "industry expert" is someone who has extensive knowledge and experience in a particular field and is able to select training data and set standards in that field.

[0133] "Training data" is a collection of data used to train an artificial intelligence to acquire the knowledge necessary to perform a specific task.

[0134] "Integrity check" refers to the process of verifying that uploaded data is in the correct format, free of missing or duplicated data, and suitable for training.

[0135] "Storage" refers to the technology used to hold uploaded data using a database or storage system.

[0136] "Training goals" refer to the specific performance or functional goals that artificial intelligence should achieve through training.

[0137] "Training standards" refer to the guidelines and rules necessary to achieve training goals, including standards for translation style and accuracy.

[0138] A "learning model" refers to an algorithm or neural network that artificial intelligence generates based on training data.

[0139] "Initialization" refers to the process of initializing the generated learning model and preparing it for training.

[0140] "Iterative learning sessions" refers to the learning process of using training data to update the learning model over multiple cycles to improve performance.

[0141] "Performance evaluation" refers to the process of calculating indicators to measure how close a learning model is to achieving its goals and analyzing the results.

[0142] "Means for notifying users" refers to the technology and methods for notifying users of performance evaluation results via a dashboard or notification function.

[0143] "Adjusting based on feedback" refers to the process of changing the parameters and training data of a learning model based on user opinions and evaluations to improve the accuracy of the model.

[0144] "Professional certification" refers to the process of officially recognizing artificial intelligence that has met its training objectives as suitable for use in practice in a particular professional field.

[0145] "Export means" refers to the technology or method for converting and outputting a certified AI model into an appropriate data format for application in a production environment.

[0146] "Applying to a production environment" refers to the process of actually implementing and using a certified AI model in a specific business or task.

[0147] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0148] Overall system configuration

[0149] Server: This is the core of the system and has the functions of receiving, storing, and processing data, as well as generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Suitable software for use includes MySQL for database management and TensorFlow or PyTorch for generating and training AI models.

[0150] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc. The interface is generally provided through a web browser.

[0151] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[0152] Program processing

[0153] The server processes the program as follows:

[0154] Data Upload and Management:

[0155] Users prepare training data related to their field of expertise. For example, if training a medical translation AI, they would select medical books, papers, medical records, and terminology dictionaries.

[0156] Through the device's web browser, the user uploads the selected training data file, for example, using an HTML5-based file upload widget.

[0157] The server receives uploaded training data, stores it in a MySQL database, and uses Python scripts to check the data for consistency and correctness.

[0158] Setting training goals and standards:

[0159] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI.

[0160] The terminal provides a form for transmitting the entered training goals and criteria to the server.

[0161] The server validates the received training goals and criteria and stores them in a database.

[0162] Generate and initialize the learning model:

[0163] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[0164] Running and monitoring the training process:

[0165] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, a Python script automatically evaluates the AI's performance and records the results.

[0166] The terminal provides a dashboard for monitoring the training process in real time and displays the evaluation results using JavaScript.

[0167] Users can check the AI's performance during training on a dashboard and provide feedback as needed. If necessary, they can input training data or instructions for adjusting the criteria into the device and send them to the server.

[0168] Rating and feedback:

[0169] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[0170] Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server.

[0171] Certification and deployment of professional AI:

[0172] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[0173] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[0174] Specific examples

[0175] As an example, we will explain a specific embodiment in which a publisher develops a medical translation AI.

[0176] 1. Data upload:

[0177] User: Selects learning data such as medical books, papers, and medical records and uploads them to the server via the device.

[0178] 2. Setting training goals and standards:

[0179] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[0180] 3. Generating and initializing the learning model:

[0181] Server: Generates and initializes AI learning models using TensorFlow and PyTorch based on medical-related data and standards.

[0182] 4. Running the training process:

[0183] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0184] 5. Evaluation and feedback:

[0185] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[0186] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[0187] 6. Certification and deployment of professional AI:

[0188] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[0189] On the device: Inform the user of the authentication information and provide an expand button.

[0190] Example prompts to input to the generative AI model

[0191] Please generate a learning model specialized for medical translation AI. Using the following data, the training goal is "improving translation accuracy" and the evaluation criteria are "accuracy of terminology and matching of context." Upload data includes medical books, papers, and medical records. Please also perform the initial setup of the training data.

[0192] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[0193] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0194] Step 1: Data upload and management

[0195] The user prepares learning data related to their field of expertise (e.g., medical books, papers, medical records, glossaries). Next, they open a web browser on their device, access the system's data upload screen, click the "Select File" button, and select the learning data file they have selected. Then, they press the "Upload" button to send the data to the server.

[0196] Input: Training data files such as medical books, papers, and medical records

[0197] Data processing and calculation: Check file format, detect missing data, delete duplicate data

[0198] Output: The training data with consistency confirmed is saved in the database.

[0199] Specific operation: The server stores the uploaded file in temporary storage, runs a Python script to check the file format, check for missing data, and remove duplicates, and then stores it in the official database.

[0200] Step 2: Setting training goals and standards

[0201] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI. Users enter the training goals and criteria in the text fields and execute the settings by pressing the save button.

[0202] Input: Training goals and standards (e.g., "Achieve medical translation accuracy of 95% or higher," "Use technical terms preferentially")

[0203] Data processing and data calculation: Validation of input data (format confirmation and content consistency confirmation)

[0204] Output: Training goals and criteria are stored in a database

[0205] Specific behavior: The server checks the received training goals and criteria with validation scripts (e.g., checking for invalid input), executes SQL queries, and records them in the database.

[0206] Step 3: Generate and initialize the learning model

[0207] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[0208] Input: Consistent training data, training criteria

[0209] Data processing and data calculation: Defining model architecture, initializing parameters, setting hyperparameters

[0210] Output: Initialized AI learning model

[0211] What happens: The server runs a Python script to define the layers of the neural network and set the initial parameters, so the learning model is ready to start training.

[0212] Step 4: Running and monitoring the training process

[0213] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, the AI's performance is automatically evaluated and the results are recorded.

[0214] Input: Consistency-checked training data, initialized AI learning model

[0215] Data processing and calculation: data batching, error calculation, backpropagation, model updating

[0216] Output: Performance evaluation results after each session

[0217] Specific operation: The server divides the training data into batches and inputs them into the model, calculates the error function for each batch, performs backpropagation to update the model, and records the training progress and evaluation results in a log file.

[0218] Step 5: Evaluate and incorporate feedback

[0219] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[0220] Input: performance results after training sessions, user feedback

[0221] Data processing and data calculation: Calculation of evaluation metrics (e.g., precision, recall, F-measure), analysis of feedback, model retraining

[0222] Output: Training results displayed in a dashboard, with an updated model incorporating feedback

[0223] Specific operation: The server executes the evaluation script to evaluate the model's performance, records the results in a database, and displays them on a dashboard. It also performs retraining and parameter adjustment based on user feedback.

[0224] Step 6: Certification and deployment of professional AI

[0225] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[0226] Input: Completely trained AI model, training result data

[0227] Data processing and calculation: Final adjustment of the model, export of the model

[0228] Output: AI models exported in a format suitable for production environments

[0229] Specific operation: The server runs an auxiliary script to evaluate the training results, and if it determines that the goal has been achieved, it exports the model in ONNX format and packages it into a Docker container.

[0230] Step 7: Notification of certification information and deployment of AI

[0231] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[0232] Input: Certified AI model, user deployment instructions

[0233] Data processing and data calculation: Notification of certification information, settings for deployment

[0234] Output: AI model deployed in production

[0235] Specific operation: The user checks the notification on the device and clicks the deploy button to deploy the AI ​​model to the production environment. The server receives the instruction and actually performs the deployment.

[0236] (Application example 1)

[0237] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0238] Conventional methods for training and evaluating AI robots specialized for specific tasks in factories are time-consuming and costly, preventing efficient automation. Furthermore, it is difficult to monitor the training process and check evaluation results in real time, making it difficult to provide appropriate feedback in a timely manner. A method for solving these problems and developing AI robots specialized in specific fields is needed.

[0239] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0240] In this invention, the server includes: means for uploading training data selected by industry experts; means for verifying the integrity of the uploaded training data and storing it; means for setting training goals and standards based on the field of expertise; means for generating and initializing a learning model for a general AI based on the set training goals and standards; means for performing repeated learning sessions on the learning model using the training data; means for evaluating the performance of the AI ​​after the learning sessions and notifying a user of the results; means for adjusting the learning model based on user feedback; means for certifying the AI ​​as a professional when training is complete and the goals are achieved; means for training an AI robot specialized in a specific task to automate work in a factory; and means for setting training data and training goals using a smartphone and checking the training process and evaluation results in real time, which not only significantly improves work efficiency in the factory and reduces time and costs, but also enables real-time monitoring of the training process and checking the evaluation results.

[0241] "General purpose artificial intelligence" refers to artificial intelligence that has a wide range of applications, not limited to specific tasks or fields.

[0242] A "specialized education platform" is a system for training and educating artificial intelligence specialized in specific fields.

[0243] An "industry expert" is an expert with advanced knowledge and experience in a particular field of expertise.

[0244] "Training data" refers to the data set used to train an artificial intelligence.

[0245] "Upload" refers to the act of transferring data from a local terminal to a server.

[0246] "Integrity" means that data is correct, consistent, and free from errors.

[0247] "Preservation" refers to the act of storing data for a long period of time.

[0248] A "training goal" is a specific goal that you want to achieve in learning artificial intelligence.

[0249] "Training standards" refer to the standards and rules that artificial intelligence must follow during the training process.

[0250] A "learning model" is an artificial intelligence algorithm built based on training data.

[0251] "Initialization" refers to the act of resetting the parameters of a learning model to their initial settings.

[0252] A "learning session" is the process by which an artificial intelligence repeatedly learns using training data.

[0253] "Performance evaluation" refers to evaluating the performance of a trained artificial intelligence.

[0254] "User" refers to a person or institution using the system.

[0255] "Feedback" refers to evaluations and comments on training results provided by users.

[0256] "Accreditation" is the act of officially recognizing a qualification or status based on specific criteria.

[0257] A "professional" is a person who has advanced knowledge and skills in a specific specialized field and is engaged in that occupation, or a system that supports such a person.

[0258] "Factory work" refers to a series of tasks and processes carried out on the factory production floor.

[0259] "Automation" is the act of enabling machines or systems to perform tasks autonomously without human intervention.

[0260] A "smartphone" is a mobile phone that has computing capabilities.

[0261] "Real time" means that events are processed and displayed as they occur.

[0262] A "dashboard" is a user interface that visually displays the status and data of a system.

[0263] The professional education platform of the present invention is designed for the purpose of training artificial intelligence robots to automate specific tasks in factories. This system mainly consists of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0264] Overall system configuration

[0265] server

[0266] The server is the core of this system and has the functions of receiving, storing, and processing data, generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Specifically, the server uses the following hardware and software:

[0267] Hardware: High performance computer, GPU (e.g. NVIDIA GPU)

[0268] Software: Python, Flask, PyTorch, Transformers library

[0269] Terminal

[0270] The terminal provides an interface for users to operate the system. Smartphones are mainly used. The terminal provides the following functions:

[0271] Uploading data

[0272] Setting training goals and standards

[0273] Monitoring the training process

[0274] Providing Feedback

[0275] User

[0276] Users are experts in factory operations and provide training data and standards, as well as feedback to the AI ​​during training.

[0277] Program processing

[0278] Data Upload and Management

[0279] User: The user prepares training data related to factory operations, such as product inspection images and sensor data.

[0280] Terminal: Upload the selected training data file to the server via the terminal.

[0281] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[0282] Setting training goals and standards

[0283] User: The user inputs the AI ​​training goal (e.g., improving the accuracy of product inspection) and training criteria (e.g., inspection speed, false positive rate) from the terminal.

[0284] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[0285] Server: Receives training goals and criteria and stores them in a database.

[0286] Generating and initializing the learning model

[0287] Server: Generates a general-purpose AI learning model based on the training data and standards received from the user. Initializes the generated learning model and performs initial settings.

[0288] Running and monitoring the training process

[0289] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., product inspection) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[0290] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[0291] Evaluation and feedback

[0292] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[0293] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0294] Certification and deployment of professional AI

[0295] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0296] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0297] Specific examples

[0298] As an example, consider an AI robot that automates the task of inspecting products on a conveyor belt in a factory. First, the user uploads the image data of the product inspection to a server using a smartphone. Next, the user inputs training goals and criteria for the AI ​​to achieve an accuracy of 95% or more in product inspection. Once the training process is complete, the user can view the evaluation results on a dashboard and provide feedback.

[0299] Example prompt sentence:

[0300] "Start the Factory Robot AI Trainer app and upload the image data of the object to be inspected. Then, enter the training goals and criteria for the AI ​​robot to achieve a product inspection accuracy of 95% or more. After training is complete, review the evaluation results and provide feedback."

[0301] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0302] Step 1:

[0303] Data upload

[0304] Users prepare training data related to factory operations (e.g., product inspection images and sensor data) and upload it to the server via their terminal. The terminal checks the file format and size of the uploaded data and transfers it to the server. The server receives the file and stores it in a database. The input is the training data, and the output is the data stored in the server's database.

[0305] Step 2:

[0306] Setting training goals and standards

[0307] The user inputs training goals (e.g., improving product inspection accuracy) and training criteria (e.g., inspection speed, false positive rate) from the terminal. The terminal sends the goals and criteria entered by the user to the server. The server receives this data and stores it in a database. The input is the goals and criteria entered by the user, and the output is the training goals and criteria stored in the server's database.

[0308] Step 3:

[0309] Generating and initializing the learning model

[0310] The server generates and initializes a general-purpose AI learning model based on the uploaded training data and the set goals and criteria. Specifically, it uses Python, PyTorch, and the Transformers library to build the learning model and set initial parameters. The input is the training data, goals, and criteria, and the output is the initialized learning model.

[0311] Step 4:

[0312] Running and monitoring the learning process

[0313] The server runs repeated learning sessions on the generated learning model. Specifically, the AI ​​performs a specific task (e.g., product inspection) using the uploaded data and records the results. After each session, the server evaluates the AI's performance and records the results. The device monitors this process in real time and displays the evaluation results on a user interface (dashboard). The inputs are the training data and the learning model, and the output is the evaluation results.

[0314] Step 5:

[0315] Evaluation and feedback

[0316] The server evaluates the AI's performance after each training session and displays the evaluation results on a dashboard. The user can check the evaluation results on the dashboard and provide feedback as needed. The server receives feedback from the user and readjusts the training model. The inputs are the evaluation results and user feedback, and the output is the adjusted training model.

[0317] Step 6:

[0318] Certification and deployment of professional AI

[0319] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI. The certified AI model is exported in an appropriate format and prepared for deployment in a production environment. The terminal notifies the user of the certification information and deploys the AI ​​in a production environment as necessary. The input is the achievement status of the training goals, and the output is the certified AI model and notification.

[0320] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0321] The specialized training platform for adapting general-purpose AI to specific specialized fields according to the present invention is composed of three elements: a server, a terminal, and a user. Furthermore, the system combines an emotion engine that recognizes user emotions and utilizes user emotional feedback in the AI ​​training process.

[0322] Overall system configuration

[0323] Server: The core of the system, it receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine.

[0324] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine.

[0325] Users: Industry experts who provide training data and standards, and emotional feedback from the emotion engine.

[0326] Emotion engine: Analyzes the user's facial expressions, voice, and input to detect emotions and reflect them in performance and feedback.

[0327] Program processing

[0328] 1. Data upload and management

[0329] User: Prepare training data related to their field of expertise. For example, if training a medical translation AI, select medical books, papers, medical records, and terminology dictionaries.

[0330] Terminal: The user uploads the selected training data file to the server through the terminal. The terminal interface assists the upload operation.

[0331] Server: Receives uploaded training data and stores it in the database. Checks the integrity and correctness of the data and stores it in the database.

[0332] 2. Setting training goals and standards

[0333] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context) from the device.

[0334] Terminal: Provides an interface for inputting training goals and criteria and transmits the settings to the server.

[0335] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[0336] 3. Generating and initializing the learning model

[0337] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[0338] 4. Running and monitoring the training process

[0339] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data. After each training session, the AI's performance is automatically evaluated and the results are recorded.

[0340] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[0341] 5. Processing emotional feedback

[0342] Device: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The emotion recognition results are sent to the server.

[0343] Server: Receives user emotion data and uses it to adjust learning models and optimize feedback responses.

[0344] 6. Evaluation and feedback

[0345] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. Readjust the learning model based on the user's emotional data.

[0346] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0347] 7. Certification and deployment of professional AI

[0348] Server: Once the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0349] Terminal: Informs users of certification information and assists them in deploying AI in production environments as needed.

[0350] Specific examples

[0351] For example, if a publisher develops a medical translation AI, the following embodiment applies.

[0352] 1. Data upload:

[0353] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[0354] 2. Setting training goals and standards:

[0355] User: Enter the goals for improving the accuracy of AI medical translation and translation style standards on the device, and send the settings to the server.

[0356] 3. Generating and initializing the learning model:

[0357] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[0358] 4. Running the training process:

[0359] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0360] 5. Processing emotional feedback:

[0361] Terminal: Recognizes the user's emotions in real time during training and sends the results to the server via the emotion engine.

[0362] Server: Leverages emotional data to adjust learning models and feedback responses.

[0363] 6. Evaluation and Feedback:

[0364] Server: Evaluates the training results and retunes the model based on user emotional feedback.

[0365] User: Modify the training data or criteria as needed and resubmit.

[0366] 7. Certification and deployment of professional AI:

[0367] Server: Executes the certification process when training objectives are achieved and deploys to the production environment.

[0368] Terminal: Notifies certification information and assists with AI deployment.

[0369] This allows the system to efficiently and effectively specialize general AI in specific fields of expertise, and comprehensively utilize user feedback to develop high-quality professional AI.

[0370] The processing flow will be explained below.

[0371] Step 1:

[0372] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0373] Step 2:

[0374] Terminal: The user uses an interface to upload the selected training data file to the server through the terminal. The user is assisted in selecting the file and pressing the upload button.

[0375] Step 3:

[0376] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the uploaded data and inspects the format and content for any problems.

[0377] Step 4:

[0378] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0379] Step 5:

[0380] Terminal: Provides an interface for inputting training goals and criteria, transmits the settings to the server, and displays a confirmation message to the user upon completion.

[0381] Step 6:

[0382] Server: Receives training goals and criteria and stores them in a database. Creates a training plan based on the received settings information.

[0383] Step 7:

[0384] Server: Generates and initializes a general-purpose AI learning model based on the learning data and training criteria received from the user. Initialization involves setting model parameters and loading initial data.

[0385] Step 8:

[0386] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[0387] Step 9:

[0388] Server: After each training session, the server automatically evaluates the AI's performance and records the results. The evaluation is based on set criteria and saved in the form of a score.

[0389] Step 10:

[0390] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to see training progress and performance.

[0391] Step 11:

[0392] Device: During training, the emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The detected emotion data is sent to the server.

[0393] Step 12:

[0394] Server: Receives and records the emotional data sent by users along with their training performance and feedback. The emotional data is used as a factor in adjusting the learning model.

[0395] Step 13:

[0396] User: Check the AI's performance during training on the dashboard and send feedback to the server from the device as needed. Feedback can be entered as text or multiple choice options.

[0397] Step 14:

[0398] Server: Retunes the learning model based on the evaluation results and sentiment data collected after each training session. Retunement includes parameter resetting and model retraining.

[0399] Step 15:

[0400] User: Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server via the terminal.

[0401] Step 16:

[0402] Server: Receives new settings from the user, adjusts the training plan again and continues the learning session, saves the settings in the database, and starts a new training session.

[0403] Step 17:

[0404] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0405] Step 18:

[0406] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​can function effectively in the field.

[0407] Example 2

[0408] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0409] A specialized training platform for effectively adapting general AI to specific specialized fields must efficiently manage the entire process from selecting training data, verifying data integrity, setting training goals and standards, running learning sessions, and evaluating and certifying performance, while incorporating and optimizing the platform by incorporating user emotional feedback.Current technology has difficulty adjusting learning models to reflect user emotions, which results in low-quality training.

[0410] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0411] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on user feedback, means for certifying the AI ​​as a professional when training is complete and the goal is achieved, and means for collecting emotional data using an emotion engine that recognizes user emotions and optimizing the learning model and feedback response based on the emotional data. This enables efficient training and adjustment of high-quality professional AI specialized in a specific field by comprehensively utilizing user emotional feedback.

[0412] "General artificial intelligence" refers to an artificial intelligence system that is not limited to a specific area of ​​expertise and can perform a wide range of tasks.

[0413] "Specialized education platform" refers to a system that provides education and training to adapt general artificial intelligence to specific specialized fields.

[0414] "Industry Expert" refers to a person or group with advanced knowledge and experience in a particular area of ​​expertise.

[0415] "Training data" refers to a dataset used to learn and train artificial intelligence.

[0416] "Integrity check" refers to the process of checking the accuracy, consistency, and correctness of uploaded training data.

[0417] "Storage" refers to storing uploaded training data in a database or storage.

[0418] "Training objectives" refer to specific performance and functional indicators that an AI system must achieve.

[0419] "Standards" refers to the standards and guidelines applied in the training process.

[0420] A "learning model" refers to the structure of algorithms and parameters created by artificial intelligence using training data to perform specific tasks.

[0421] "Initialization" refers to the process of setting the parameters of a learning model to an initial state.

[0422] "Repeated learning sessions" refers to repeating the learning process multiple times using training data.

[0423] "Performance evaluation" refers to the process of measuring the performance and accuracy of an artificial intelligence system and analyzing the results.

[0424] "User" refers to the person who operates the professional education platform, provides training data, and sets training goals.

[0425] "Feedback" refers to the evaluation and suggestions provided by the user regarding the training results.

[0426] An "emotion engine" refers to a system that recognizes and analyzes emotions from a user's facial expressions, voice, and input content.

[0427] "Emotion data" refers to the user's emotion information detected by the emotion engine.

[0428] "Optimization" refers to the process of improving the performance of a learning model or system based on emotional data and feedback.

[0429] "Professional AI" refers to an artificial intelligence system that has advanced knowledge and skills in a specific specialized field and is useful in practical work.

[0430] "Certification" refers to the process of officially recognizing that training objectives have been achieved.

[0431] The specialized education platform of the present invention, which adapts a general-purpose artificial intelligence to a specific specialized field, is composed of three elements: a server, a terminal, and a user. The core of this system is the server, which receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine. Specific embodiments are described in detail below.

[0432] Overall structure

[0433] Server: The central component of the system, responsible for receiving, storing, and processing data, generating and training learning models, and performing all processes, including checking the integrity of training data, setting training goals, generating learning models, conducting repeated training sessions, evaluating performance, incorporating user feedback, and aggregating and utilizing emotional data.

[0434] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine. Users upload training data and input settings through the terminal.

[0435] User: Prepares training data in the field of expertise, sets training goals and standards, checks training status and results, provides feedback, and provides emotional feedback through the emotion engine.

[0436] Hardware / Software Used

[0437] Database: A database system for storing learning data and the training goals and criteria that are set.

[0438] Emotion engine: A software module that analyzes the user's facial expressions, voice, and input content to detect emotions.

[0439] Dashboard: A graphical user interface (GUI) for displaying the training process and the results of evaluating the AI's performance in real time.

[0440] Specific examples

[0441] Developing AI for medical translation

[0442] As an example, an embodiment in which a publisher develops a medical translation AI will be described.

[0443] 1. Data upload

[0444] User: Selects relevant materials such as medical books, papers, and medical records and uploads them to the server via the terminal.

[0445] Terminal: Provides an interface for selecting files and showing the progress of the upload.

[0446] Server: Receives the uploaded training data, checks its integrity, and stores it in the database.

[0447] 2. Setting training goals and standards

[0448] Users: Aiming to improve the accuracy of medical translations, they set standards such as consistency in translation style and terminology.

[0449] Terminal: Provides an interface for setting input and sends user input to the server.

[0450] Server: Receives the set training goals and criteria, stores them in a database, and then creates a training plan.

[0451] 3. Generating and initializing the learning model

[0452] Server: Generates and initializes the medical translation AI learning model based on medical-related data and standards, using algorithms such as the Transformer model and LSTM.

[0453] 4. Running the training process

[0454] Server: Using the training data, it runs iterative training sessions to train the AI ​​on the task of translating medical documents and optimize the model parameters.

[0455] 5. Processing emotional feedback

[0456] Device: Analyzes the user's facial expressions and voice in real time during training and collects emotional data through an emotion engine.

[0457] Server: Receives emotion data and uses it to adjust the learning model.

[0458] 6. Ratings and Feedback

[0459] Server: Evaluates the AI's performance after each training session and notifies the user of the results.

[0460] User: Modify the training data and configuration criteria as needed and submit the new configuration to the server.

[0461] 7. Certification and Deployment

[0462] Server: Once training objectives are met, certify the AI ​​as a professional and prepare it for deployment in a production environment.

[0463] Terminal: Provides an interface that notifies the user of certified information and assists in deployment in the production environment.

[0464] Prompt Sentence Examples

[0465] "Upload medical books and medical papers to start the training process of the AI ​​learning model. The training goal is to improve the accuracy of translations in the medical field."

[0466] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[0467] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0468] Step 1: Upload and manage your data

[0469] Input: Training data related to the specialty (e.g., medical books, papers, medical records).

[0470] User action: The user prepares a pre-selected training data file. For example, the user saves medical books or papers on their device.

[0471] Device operation: The user operates the device to upload the training data file to the server. The device displays a file selection screen and shows the upload progress of the file selected by the user.

[0472] Server operation: The server receives the uploaded file and checks the data for consistency (no duplicates or format inconsistencies). Once the validation is complete, the data is stored in the database.

[0473] Output: The validated training data is saved in the database.

[0474] Step 2: Setting training goals and standards

[0475] Input: Training goals and criteria (e.g., translation accuracy, style criteria).

[0476] User action: The user inputs the AI ​​training goal (e.g., "Improve the accuracy of medical translation") and training criteria (e.g., "Comply with the latest medical terminology dictionary") through the terminal.

[0477] Terminal Operation: The terminal provides an interface for inputting training goals and criteria and transmits the user's input to the server.

[0478] Server Operation: The server receives the input training goals and criteria and stores them in a database.

[0479] Output: The saved training goals and criteria are present in the database.

[0480] Step 3: Generate and initialize the learning model

[0481] Input: Training data and training target,criteria.

[0482] Server operation: The server retrieves training data, training goals, and criteria from the database, and generates an AI learning model based on them. The algorithms used may be Transformer or LSTM. The server initializes the model and sets its parameters.

[0483] Output: A trained model with initial setup completed.

[0484] Step 4: Running and monitoring the training process

[0485] Input: Initialized learning model, training data.

[0486] Server operation: The server inputs training data into the learning model and runs iterative training sessions. In each session, the AI ​​performs a specified task (e.g., translating medical documents) and optimizes the model's parameters.

[0487] Terminal operation: The terminal displays a dashboard for monitoring the training progress in real time. Users can check the progress through the terminal.

[0488] Server behavior: After a training session, the server evaluates the model's performance and records the results.

[0489] Output: Performance evaluation results and recorded data.

[0490] Step 5: Processing emotional feedback

[0491] Input: Emotional feedback (facial expressions, voice, input).

[0492] Device operation: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions.

[0493] Device operation: The detected emotion data is sent to the server.

[0494] Server operation: The server receives the user's emotional data and uses it to readjust the learning model and optimize the feedback response. For example, if the user is in a high-stress state, it will adjust the training pace.

[0495] Output: The adjusted learning model.

[0496] Step 6: Evaluate and incorporate feedback

[0497] Input: Performance evaluation results, feedback.

[0498] Server behavior: Evaluate the AI's performance after each training session. The results are displayed on a dashboard and notified to the user.

[0499] User action: The user checks the evaluation results, modifies the training data and configuration criteria as needed, and submits the new configuration to the server.

[0500] Server action: Retune the learning model based on feedback.

[0501] Output: An improved learning model.

[0502] Step 7: Certify and deploy professional AI

[0503] Input: Training goal achievement status.

[0504] Server operation: If the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI. The server exports the certified AI model in an appropriate format (e.g., API or software module) and prepares it for deployment in a production environment.

[0505] Terminal operation: Informs the user of certification information and provides an interface to assist with the steps to deploy AI in a production environment.

[0506] Output: Professional AI deployed in a production environment.

[0507] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[0508] (Application example 2)

[0509] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0510] Currently, there are educational platforms for adapting AGI to specific specialized fields, but few systems incorporate emotional feedback, limiting their effectiveness and efficiency. In particular, in practical environments such as logistics centers, workers' emotions often have a significant impact on work efficiency. Therefore, there is a need to develop a system that combines AGI training with workers' emotional feedback to optimize work flows and improve the accuracy of learning models.

[0511] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0512] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general-purpose AI based on the set training goals and standards, means for performing repetitive learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning sessions and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is complete and the goals are achieved, and means for analyzing worker emotional feedback in real time and using the results to adjust the learning model and optimize work processes. This enables real-time work adjustments in accordance with worker emotions at logistics centers, thereby improving the accuracy of the general-purpose AI and significantly improving work efficiency.

[0513] Definition of Terms

[0514] "General artificial intelligence" is an artificial intelligence system that can adapt to a wide range of tasks and fields.

[0515] A "specialized education platform" is an educational system for providing specialized training in a specific field of expertise.

[0516] "Emotional feedback" refers to feedback data based on a user's emotional state.

[0517] An "emotion engine" is software or hardware that analyzes a user's facial expressions and voice to detect emotions.

[0518] A "logistics center" is a facility that stores, sorts, and distributes goods.

[0519] "Training data" is a data set used to train an artificial intelligence learning model.

[0520] A "learning model" is an algorithmic structure that is generated using training data to perform a specific task.

[0521] "Real-time" refers to user operations and data processing being carried out immediately without delay.

[0522] "Business process optimization" means improving processes to maximize the efficiency and effectiveness of business operations.

[0523] A "repeated learning session" is the process of repeatedly performing the same or similar tasks to train a learning model.

[0524] A "dashboard" is a graphical interface that visually displays the status and performance of a system.

[0525] "Feedback" refers to evaluations and opinions on the performance and results of a system.

[0526] "Professional certification" means officially recognizing someone as having particular skills or knowledge.

[0527] MODE FOR CARRYING OUT THE INVENTION

[0528] To implement this invention, the following system configuration is required: The system is made up of three elements: a server, a terminal, and a user. The roles and functions of each will be described in detail below.

[0529] Components

[0530] 1. Server:

[0531] Function: The server is the core of the system, receiving, storing, and processing data, generating and training learning models, evaluating them, and processing the emotion engine.

[0532] Software used: Database management software, Python

[0533] For example: The server receives training data uploaded by industry experts and stores it in a database. It then generates an AI learning model based on training goals and criteria and runs repeated learning sessions. It evaluates the performance of the learning model and sends the results to the device. It also receives emotional data sent from the device and adjusts the learning model and business process based on that data.

[0534] 2. Terminal:

[0535] Function: Provides an interface for users to operate the system, and has emotion recognition functionality using an emotion engine.

[0536] Hardware used: Smart glasses

[0537] Software used: Emotion engine, dashboard viewer

[0538] Example: A worker wears smart glasses, and the emotion engine analyzes his facial expressions and voice in real time while he works. The user sets training goals and standards through the device, uploads data, and enters feedback. In addition, the evaluation results of the learning model are displayed on a dashboard, allowing the user to check the current training progress.

[0539] 3. User:

[0540] Functions: Industry experts, providing training data and standards, emotional feedback through emotion engine.

[0541] Example: As a work manager at a logistics center, the user uploads work-related logs and data to the server, sets training goals and standards for the AI, and sends feedback and optimization of work flows to the server based on emotional data acquired during work.

[0542] Add examples to your description

[0543] Examples:

[0544] Consider a logistics center where a worker is wearing smart glasses and working in a picking area. For example, if the emotion engine detects fatigue while the worker is heading to the next picking area, it sends the result to the server. The server immediately reduces the worker's tasks temporarily and automatically assigns the tasks to a robot in the picking area. In this way, it is possible to maintain work efficiency while reducing the burden on the worker.

[0545] Example prompt sentence:

[0546] "Generate the behavior of a smart logistics system based on the following logistics center scenario. In the logistics center, workers wear smart glasses and work collaboratively with robots. Describe in detail the behavior of the system that efficiently assigns tasks based on the workers' emotional feedback and supports the work with robots."

[0547] This will enable logistics centers to adjust operations in real time according to the emotions of workers, improving the accuracy of general-purpose artificial intelligence and significantly improving operational efficiency.

[0548] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0549] Program processing steps

[0550] Step 1:

[0551] The server receives training data selected by industry experts. Users upload the training data through their devices and store it in the database after checking the integrity of the entered data. Data integrity checking includes checking the data format and error handling.

[0552] Input: Training data (e.g., logistics-related data)

[0553] Output: Consistent training data stored in a database

[0554] Step 2:

[0555] Users input training goals and standards based on their field of expertise on their terminal and send them to the server, which then sets the goals and standards that the training model must achieve.

[0556] Input: Training goals and standards (e.g., improving picking efficiency)

[0557] Output: Training goals and criteria stored in a database

[0558] Step 3:

[0559] The server generates and initializes a general AI learning model based on the uploaded training data and the set training goals and criteria. In particular, a learning model specialized for logistics operations is generated here.

[0560] Input: Stored training data, training goals and criteria

[0561] Output: Initialized training model

[0562] Step 4:

[0563] The server runs repeated training sessions on the learning model, having the model perform business tasks based on the training data, and evaluates its performance. After each session, the evaluation results are recorded in a database and sent to the device.

[0564] Input: Learning model, training data

[0565] Output: Evaluation results, updated learning model

[0566] Step 5:

[0567] The device analyzes the user's facial expressions and voice in real time through the emotion engine, generating emotion data, which is then sent to the server.

[0568] Input: User's facial expression, voice

[0569] Output: Emotion data

[0570] Step 6:

[0571] The server receives the acquired emotion data and reflects it in adjusting the performance of the learning model and the workflow. Specifically, it takes operational optimization measures such as reallocating tasks according to the worker's emotions.

[0572] Input: Emotion data, learning model

[0573] Output: Optimized workflow, adjusted learning model

[0574] Step 7:

[0575] Once training is complete and the objectives are met, the server executes the process of certifying the AI ​​as a professional, and the certified AI is deployed in a practical environment such as a logistics center.

[0576] Input: Results of completed training sessions, evaluation data

[0577] Output: Certified artificial intelligence, ready for production deployment

[0578] This allows the server, terminal, and user to work together to realize learning and optimization of general-purpose artificial intelligence based on emotional feedback, enabling efficient business operations at logistics centers.

[0579] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0580] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0581] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0582] [Second embodiment]

[0583] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0584] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0585] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0586] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0587] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0588] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0589] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0590] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0591] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0592] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0593] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0594] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0595] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0596] Overall system configuration

[0597] Server: The core of the system, it receives, stores, processes data, generates, trains, and evaluates learning models. The server manages multiple AI learning models and serves as a specialized education platform.

[0598] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc.

[0599] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[0600] Program processing

[0601] 1. Data upload and management

[0602] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0603] Terminal: The user uploads the selected training data file to the server via the terminal.

[0604] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[0605] 2. Setting training goals and standards

[0606] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0607] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[0608] Server: Receives training goals and criteria and stores them in a database.

[0609] 3. Generating and initializing the learning model

[0610] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the generated learning model and performs initial settings (e.g., setting initial parameter values, setting hyperparameters for training).

[0611] 4. Running and monitoring the training process

[0612] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[0613] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[0614] User: View the AI's performance during training on a dashboard and provide feedback as needed. If the training data or criteria need to be adjusted, send new instructions to the server.

[0615] 5. Evaluation and feedback

[0616] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[0617] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0618] 6. Certification and deployment of professional AI

[0619] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0620] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0621] Specific examples

[0622] For example, if a publisher develops a medical translation AI, the following embodiment applies:

[0623] 1. Data upload:

[0624] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[0625] 2. Setting training goals and standards:

[0626] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[0627] 3. Generating and initializing the learning model:

[0628] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[0629] 4. Running the training process:

[0630] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0631] 5. Evaluation and feedback:

[0632] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[0633] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[0634] 6. Certification and deployment of professional AI:

[0635] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[0636] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0637] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[0638] The processing flow will be explained below.

[0639] Step 1:

[0640] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0641] Step 2:

[0642] Device: The user uploads the selected training data file to the server via the device. An upload UI is provided to assist with file selection and transmission.

[0643] Step 3:

[0644] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the data and automatically inspects the format and content of the data before storing it.

[0645] Step 4:

[0646] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0647] Step 5:

[0648] Terminal: Provides a UI for inputting training goals and criteria, sends the settings to the server, and displays a confirmation message to the user when the sending operation is complete.

[0649] Step 6:

[0650] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[0651] Step 7:

[0652] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[0653] Step 8:

[0654] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[0655] Step 9:

[0656] Server: Automatically evaluates the AI's performance after each training session, records the results, scores them based on the evaluation criteria, and stores them in a database.

[0657] Step 10:

[0658] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[0659] Step 11:

[0660] User: Check the AI's performance during training on the dashboard and provide feedback as needed. Feedback is sent from the device to the server.

[0661] Step 12:

[0662] Server: After each training session, the server evaluates the AI's performance and uses the feedback to retune the learning model, resetting parameters as needed.

[0663] Step 13:

[0664] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0665] Step 14:

[0666] Server: Receives new settings from the user, adjusts the training plan again, and continues the learning session.

[0667] Step 15:

[0668] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0669] Step 16:

[0670] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​works effectively in the field.

[0671] Example 1

[0672] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0673] Adapting general-purpose AI to specific specialized fields requires effective management of diverse training data and the generation of highly accurate learning models. However, conventional systems have not adequately verified the consistency of specialized training data, monitored the training process in real time, or applied the data to a practical environment after the training goal has been achieved. As a result, it has been difficult to efficiently and effectively develop specialized AI and deploy it in practice.

[0674] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0675] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model of the general purpose AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is completed and the goal is achieved, and means for exporting the generated professional AI model for application to a practical environment. This makes it possible to efficiently and effectively adapt the general purpose AI to a specific field of expertise and to develop and deploy professional AI that is suited to practical use.

[0676] An "industry expert" is someone who has extensive knowledge and experience in a particular field and is able to select training data and set standards in that field.

[0677] "Training data" is a collection of data used to train an artificial intelligence to acquire the knowledge necessary to perform a specific task.

[0678] "Integrity check" refers to the process of verifying that uploaded data is in the correct format, free of missing or duplicated data, and suitable for training.

[0679] "Storage" refers to the technology used to hold uploaded data using a database or storage system.

[0680] "Training goals" refer to the specific performance or functional goals that artificial intelligence should achieve through training.

[0681] "Training standards" refer to the guidelines and rules necessary to achieve training goals, including standards for translation style and accuracy.

[0682] A "learning model" refers to an algorithm or neural network that artificial intelligence generates based on training data.

[0683] "Initialization" refers to the process of initializing the generated learning model and preparing it for training.

[0684] "Iterative learning sessions" refers to the learning process of using training data to update the learning model over multiple cycles to improve performance.

[0685] "Performance evaluation" refers to the process of calculating indicators to measure how close a learning model is to achieving its goals and analyzing the results.

[0686] "Means for notifying users" refers to the technology and methods for notifying users of performance evaluation results via a dashboard or notification function.

[0687] "Adjusting based on feedback" refers to the process of changing the parameters and training data of a learning model based on user opinions and evaluations to improve the accuracy of the model.

[0688] "Professional certification" refers to the process of officially recognizing artificial intelligence that has met its training objectives as suitable for use in practice in a particular professional field.

[0689] "Export means" refers to the technology or method for converting and outputting a certified AI model into an appropriate data format for application in a production environment.

[0690] "Applying to a production environment" refers to the process of actually implementing and using a certified AI model in a specific business or task.

[0691] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0692] Overall system configuration

[0693] Server: This is the core of the system and has the functions of receiving, storing, and processing data, as well as generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Suitable software for use includes MySQL for database management and TensorFlow or PyTorch for generating and training AI models.

[0694] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc. The interface is generally provided through a web browser.

[0695] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[0696] Program processing

[0697] The server processes the program as follows:

[0698] Data Upload and Management:

[0699] Users prepare training data related to their field of expertise. For example, if training a medical translation AI, they would select medical books, papers, medical records, and terminology dictionaries.

[0700] Through the device's web browser, the user uploads the selected training data file, for example, using an HTML5-based file upload widget.

[0701] The server receives uploaded training data, stores it in a MySQL database, and uses Python scripts to check the data for consistency and correctness.

[0702] Setting training goals and standards:

[0703] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI.

[0704] The terminal provides a form for transmitting the entered training goals and criteria to the server.

[0705] The server validates the received training goals and criteria and stores them in a database.

[0706] Generate and initialize the learning model:

[0707] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[0708] Running and monitoring the training process:

[0709] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, a Python script automatically evaluates the AI's performance and records the results.

[0710] The terminal provides a dashboard for monitoring the training process in real time and displays the evaluation results using JavaScript.

[0711] Users can check the AI's performance during training on a dashboard and provide feedback as needed. If necessary, they can input training data or instructions for adjusting the criteria into the device and send them to the server.

[0712] Rating and feedback:

[0713] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[0714] Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server.

[0715] Certification and deployment of professional AI:

[0716] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[0717] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[0718] Specific examples

[0719] As an example, we will explain a specific embodiment in which a publisher develops a medical translation AI.

[0720] 1. Data upload:

[0721] User: Selects learning data such as medical books, papers, and medical records and uploads them to the server via the device.

[0722] 2. Setting training goals and standards:

[0723] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[0724] 3. Generating and initializing the learning model:

[0725] Server: Generates and initializes AI learning models using TensorFlow and PyTorch based on medical-related data and standards.

[0726] 4. Running the training process:

[0727] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0728] 5. Evaluation and feedback:

[0729] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[0730] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[0731] 6. Certification and deployment of professional AI:

[0732] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[0733] On the device: Inform the user of the authentication information and provide an expand button.

[0734] Example prompts to input to the generative AI model

[0735] Please generate a learning model specialized for medical translation AI. Using the following data, the training goal is "improving translation accuracy" and the evaluation criteria are "accuracy of terminology and matching of context." Upload data includes medical books, papers, and medical records. Please also perform the initial setup of the training data.

[0736] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[0737] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0738] Step 1: Data upload and management

[0739] The user prepares learning data related to their field of expertise (e.g., medical books, papers, medical records, glossaries). Next, they open a web browser on their device, access the system's data upload screen, click the "Select File" button, and select the learning data file they have selected. Then, they press the "Upload" button to send the data to the server.

[0740] Input: Training data files such as medical books, papers, and medical records

[0741] Data processing and calculation: Check file format, detect missing data, delete duplicate data

[0742] Output: The training data with consistency confirmed is saved in the database.

[0743] Specific operation: The server stores the uploaded file in temporary storage, runs a Python script to check the file format, check for missing data, and remove duplicates, and then stores it in the official database.

[0744] Step 2: Setting training goals and standards

[0745] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI. Users enter the training goals and criteria in the text fields and execute the settings by pressing the save button.

[0746] Input: Training goals and standards (e.g., "Achieve medical translation accuracy of 95% or higher," "Use technical terms preferentially")

[0747] Data processing and data calculation: Validation of input data (format confirmation and content consistency confirmation)

[0748] Output: Training goals and criteria are stored in a database

[0749] Specific behavior: The server checks the received training goals and criteria with validation scripts (e.g., checking for invalid input), executes SQL queries, and records them in the database.

[0750] Step 3: Generate and initialize the learning model

[0751] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[0752] Input: Consistent training data, training criteria

[0753] Data processing and data calculation: Defining model architecture, initializing parameters, setting hyperparameters

[0754] Output: Initialized AI learning model

[0755] What happens: The server runs a Python script to define the layers of the neural network and set the initial parameters, so the learning model is ready to start training.

[0756] Step 4: Running and monitoring the training process

[0757] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, the AI's performance is automatically evaluated and the results are recorded.

[0758] Input: Consistency-checked training data, initialized AI learning model

[0759] Data processing and calculation: data batching, error calculation, backpropagation, model updating

[0760] Output: Performance evaluation results after each session

[0761] Specific operation: The server divides the training data into batches and inputs them into the model, calculates the error function for each batch, performs backpropagation to update the model, and records the training progress and evaluation results in a log file.

[0762] Step 5: Evaluate and incorporate feedback

[0763] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[0764] Input: performance results after training sessions, user feedback

[0765] Data processing and data calculation: Calculation of evaluation metrics (e.g., precision, recall, F-measure), analysis of feedback, model retraining

[0766] Output: Training results displayed in a dashboard, with an updated model incorporating feedback

[0767] Specific operation: The server executes the evaluation script to evaluate the model's performance, records the results in a database, and displays them on a dashboard. It also performs retraining and parameter adjustment based on user feedback.

[0768] Step 6: Certification and deployment of professional AI

[0769] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[0770] Input: Completely trained AI model, training result data

[0771] Data processing and calculation: Final adjustment of the model, export of the model

[0772] Output: AI models exported in a format suitable for production environments

[0773] Specific operation: The server runs an auxiliary script to evaluate the training results, and if it determines that the goal has been achieved, it exports the model in ONNX format and packages it into a Docker container.

[0774] Step 7: Notification of certification information and deployment of AI

[0775] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[0776] Input: Certified AI model, user deployment instructions

[0777] Data processing and data calculation: Notification of certification information, settings for deployment

[0778] Output: AI model deployed in production

[0779] Specific operation: The user checks the notification on the device and clicks the deploy button to deploy the AI ​​model to the production environment. The server receives the instruction and actually performs the deployment.

[0780] (Application example 1)

[0781] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0782] Conventional methods for training and evaluating AI robots specialized for specific tasks in factories are time-consuming and costly, preventing efficient automation. Furthermore, it is difficult to monitor the training process and check evaluation results in real time, making it difficult to provide appropriate feedback in a timely manner. A method for solving these problems and developing AI robots specialized in specific fields is needed.

[0783] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0784] In this invention, the server includes: means for uploading training data selected by industry experts; means for verifying the integrity of the uploaded training data and storing it; means for setting training goals and standards based on the field of expertise; means for generating and initializing a learning model for a general AI based on the set training goals and standards; means for performing repeated learning sessions on the learning model using the training data; means for evaluating the performance of the AI ​​after the learning sessions and notifying a user of the results; means for adjusting the learning model based on user feedback; means for certifying the AI ​​as a professional when training is complete and the goals are achieved; means for training an AI robot specialized in a specific task to automate work in a factory; and means for setting training data and training goals using a smartphone and checking the training process and evaluation results in real time, which not only significantly improves work efficiency in the factory and reduces time and costs, but also enables real-time monitoring of the training process and checking the evaluation results.

[0785] "General purpose artificial intelligence" refers to artificial intelligence that has a wide range of applications, not limited to specific tasks or fields.

[0786] A "specialized education platform" is a system for training and educating artificial intelligence specialized in specific fields.

[0787] An "industry expert" is an expert with advanced knowledge and experience in a particular field of expertise.

[0788] "Training data" refers to the data set used to train an artificial intelligence.

[0789] "Upload" refers to the act of transferring data from a local terminal to a server.

[0790] "Integrity" means that data is correct, consistent, and free from errors.

[0791] "Preservation" refers to the act of storing data for a long period of time.

[0792] A "training goal" is a specific goal that you want to achieve in learning artificial intelligence.

[0793] "Training standards" refer to the standards and rules that artificial intelligence must follow during the training process.

[0794] A "learning model" is an artificial intelligence algorithm built based on training data.

[0795] "Initialization" refers to the act of resetting the parameters of a learning model to their initial settings.

[0796] A "learning session" is the process by which an artificial intelligence repeatedly learns using training data.

[0797] "Performance evaluation" refers to evaluating the performance of a trained artificial intelligence.

[0798] "User" refers to a person or institution using the system.

[0799] "Feedback" refers to evaluations and comments on training results provided by users.

[0800] "Accreditation" is the act of officially recognizing a qualification or status based on specific criteria.

[0801] A "professional" is a person who has advanced knowledge and skills in a specific specialized field and is engaged in that occupation, or a system that supports such a person.

[0802] "Factory work" refers to a series of tasks and processes carried out on the factory production floor.

[0803] "Automation" is the act of enabling machines or systems to perform tasks autonomously without human intervention.

[0804] A "smartphone" is a mobile phone that has computing capabilities.

[0805] "Real time" means that events are processed and displayed as they occur.

[0806] A "dashboard" is a user interface that visually displays the status and data of a system.

[0807] The professional education platform of the present invention is designed for the purpose of training artificial intelligence robots to automate specific tasks in factories. This system mainly consists of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[0808] Overall system configuration

[0809] server

[0810] The server is the core of this system and has the functions of receiving, storing, and processing data, generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Specifically, the server uses the following hardware and software:

[0811] Hardware: High performance computer, GPU (e.g. NVIDIA GPU)

[0812] Software: Python, Flask, PyTorch, Transformers library

[0813] Terminal

[0814] The terminal provides an interface for users to operate the system. Smartphones are mainly used. The terminal provides the following functions:

[0815] Uploading data

[0816] Setting training goals and standards

[0817] Monitoring the training process

[0818] Providing Feedback

[0819] User

[0820] Users are experts in factory operations and provide training data and standards, as well as feedback to the AI ​​during training.

[0821] Program processing

[0822] Data Upload and Management

[0823] User: The user prepares training data related to factory operations, such as product inspection images and sensor data.

[0824] Terminal: Upload the selected training data file to the server via the terminal.

[0825] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[0826] Setting training goals and standards

[0827] User: The user inputs the AI ​​training goal (e.g., improving the accuracy of product inspection) and training criteria (e.g., inspection speed, false positive rate) from the terminal.

[0828] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[0829] Server: Receives training goals and criteria and stores them in a database.

[0830] Generating and initializing the learning model

[0831] Server: Generates a general-purpose AI learning model based on the training data and standards received from the user. Initializes the generated learning model and performs initial settings.

[0832] Running and monitoring the training process

[0833] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., product inspection) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[0834] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[0835] Evaluation and feedback

[0836] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[0837] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0838] Certification and deployment of professional AI

[0839] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0840] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[0841] Specific examples

[0842] As an example, consider an AI robot that automates the task of inspecting products on a conveyor belt in a factory. First, the user uploads the image data of the product inspection to a server using a smartphone. Next, the user inputs training goals and criteria for the AI ​​to achieve an accuracy of 95% or more in product inspection. Once the training process is complete, the user can view the evaluation results on a dashboard and provide feedback.

[0843] Example prompt sentence:

[0844] "Start the Factory Robot AI Trainer app and upload the image data of the object to be inspected. Then, enter the training goals and criteria for the AI ​​robot to achieve a product inspection accuracy of 95% or more. After training is complete, review the evaluation results and provide feedback."

[0845] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0846] Step 1:

[0847] Data upload

[0848] Users prepare training data related to factory operations (e.g., product inspection images and sensor data) and upload it to the server via their terminal. The terminal checks the file format and size of the uploaded data and transfers it to the server. The server receives the file and stores it in a database. The input is the training data, and the output is the data stored in the server's database.

[0849] Step 2:

[0850] Setting training goals and standards

[0851] The user inputs training goals (e.g., improving product inspection accuracy) and training criteria (e.g., inspection speed, false positive rate) from the terminal. The terminal sends the goals and criteria entered by the user to the server. The server receives this data and stores it in a database. The input is the goals and criteria entered by the user, and the output is the training goals and criteria stored in the server's database.

[0852] Step 3:

[0853] Generating and initializing the learning model

[0854] The server generates and initializes a general-purpose AI learning model based on the uploaded training data and the set goals and criteria. Specifically, it uses Python, PyTorch, and the Transformers library to build the learning model and set initial parameters. The input is the training data, goals, and criteria, and the output is the initialized learning model.

[0855] Step 4:

[0856] Running and monitoring the learning process

[0857] The server runs repeated learning sessions on the generated learning model. Specifically, the AI ​​performs a specific task (e.g., product inspection) using the uploaded data and records the results. After each session, the server evaluates the AI's performance and records the results. The device monitors this process in real time and displays the evaluation results on a user interface (dashboard). The inputs are the training data and the learning model, and the output is the evaluation results.

[0858] Step 5:

[0859] Evaluation and feedback

[0860] The server evaluates the AI's performance after each training session and displays the evaluation results on a dashboard. The user can check the evaluation results on the dashboard and provide feedback as needed. The server receives feedback from the user and readjusts the training model. The inputs are the evaluation results and user feedback, and the output is the adjusted training model.

[0861] Step 6:

[0862] Certification and deployment of professional AI

[0863] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI. The certified AI model is exported in an appropriate format and prepared for deployment in a production environment. The terminal notifies the user of the certification information and deploys the AI ​​in a production environment as necessary. The input is the achievement status of the training goals, and the output is the certified AI model and notification.

[0864] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0865] The specialized training platform for adapting general-purpose AI to specific specialized fields according to the present invention is composed of three elements: a server, a terminal, and a user. Furthermore, the system combines an emotion engine that recognizes user emotions and utilizes user emotional feedback in the AI ​​training process.

[0866] Overall system configuration

[0867] Server: The core of the system, it receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine.

[0868] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine.

[0869] Users: Industry experts who provide training data and standards, and emotional feedback from the emotion engine.

[0870] Emotion engine: Analyzes the user's facial expressions, voice, and input to detect emotions and reflect them in performance and feedback.

[0871] Program processing

[0872] 1. Data upload and management

[0873] User: Prepare training data related to their field of expertise. For example, if training a medical translation AI, select medical books, papers, medical records, and terminology dictionaries.

[0874] Terminal: The user uploads the selected training data file to the server through the terminal. The terminal interface assists the upload operation.

[0875] Server: Receives uploaded training data and stores it in the database. Checks the integrity and correctness of the data and stores it in the database.

[0876] 2. Setting training goals and standards

[0877] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context) from the device.

[0878] Terminal: Provides an interface for inputting training goals and criteria and transmits the settings to the server.

[0879] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[0880] 3. Generating and initializing the learning model

[0881] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[0882] 4. Running and monitoring the training process

[0883] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data. After each training session, the AI's performance is automatically evaluated and the results are recorded.

[0884] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[0885] 5. Processing emotional feedback

[0886] Device: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The emotion recognition results are sent to the server.

[0887] Server: Receives user emotion data and uses it to adjust learning models and optimize feedback responses.

[0888] 6. Evaluation and feedback

[0889] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. Readjust the learning model based on the user's emotional data.

[0890] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[0891] 7. Certification and deployment of professional AI

[0892] Server: Once the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0893] Terminal: Informs users of certification information and assists them in deploying AI in production environments as needed.

[0894] Specific examples

[0895] For example, if a publisher develops a medical translation AI, the following embodiment applies.

[0896] 1. Data upload:

[0897] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[0898] 2. Setting training goals and standards:

[0899] User: Enter the goals for improving the accuracy of AI medical translation and translation style standards on the device, and send the settings to the server.

[0900] 3. Generating and initializing the learning model:

[0901] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[0902] 4. Running the training process:

[0903] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[0904] 5. Processing emotional feedback:

[0905] Terminal: Recognizes the user's emotions in real time during training and sends the results to the server via the emotion engine.

[0906] Server: Leverages emotional data to adjust learning models and feedback responses.

[0907] 6. Evaluation and Feedback:

[0908] Server: Evaluates the training results and retunes the model based on user emotional feedback.

[0909] User: Modify the training data or criteria as needed and resubmit.

[0910] 7. Certification and deployment of professional AI:

[0911] Server: Executes the certification process when training objectives are achieved and deploys to the production environment.

[0912] Terminal: Notifies certification information and assists with AI deployment.

[0913] This allows the system to efficiently and effectively specialize general AI in specific fields of expertise, and comprehensively utilize user feedback to develop high-quality professional AI.

[0914] The processing flow will be explained below.

[0915] Step 1:

[0916] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[0917] Step 2:

[0918] Terminal: The user uses an interface to upload the selected training data file to the server through the terminal. The user is assisted in selecting the file and pressing the upload button.

[0919] Step 3:

[0920] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the uploaded data and inspects the format and content for any problems.

[0921] Step 4:

[0922] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[0923] Step 5:

[0924] Terminal: Provides an interface for inputting training goals and criteria, transmits the settings to the server, and displays a confirmation message to the user upon completion.

[0925] Step 6:

[0926] Server: Receives training goals and criteria and stores them in a database. Creates a training plan based on the received settings information.

[0927] Step 7:

[0928] Server: Generates and initializes a general-purpose AI learning model based on the learning data and training criteria received from the user. Initialization involves setting model parameters and loading initial data.

[0929] Step 8:

[0930] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[0931] Step 9:

[0932] Server: After each training session, the server automatically evaluates the AI's performance and records the results. The evaluation is based on set criteria and saved in the form of a score.

[0933] Step 10:

[0934] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to see training progress and performance.

[0935] Step 11:

[0936] Device: During training, the emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The detected emotion data is sent to the server.

[0937] Step 12:

[0938] Server: Receives and records the emotional data sent by users along with their training performance and feedback. The emotional data is used as a factor in adjusting the learning model.

[0939] Step 13:

[0940] User: Check the AI's performance during training on the dashboard and send feedback to the server from the device as needed. Feedback can be entered as text or multiple choice options.

[0941] Step 14:

[0942] Server: Retunes the learning model based on the evaluation results and sentiment data collected after each training session. Retunement includes parameter resetting and model retraining.

[0943] Step 15:

[0944] User: Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server via the terminal.

[0945] Step 16:

[0946] Server: Receives new settings from the user, adjusts the training plan again and continues the learning session, saves the settings in the database, and starts a new training session.

[0947] Step 17:

[0948] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[0949] Step 18:

[0950] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​can function effectively in the field.

[0951] Example 2

[0952] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0953] A specialized training platform for effectively adapting general AI to specific specialized fields must efficiently manage the entire process from selecting training data, verifying data integrity, setting training goals and standards, running learning sessions, and evaluating and certifying performance, while incorporating and optimizing the platform by incorporating user emotional feedback.Current technology has difficulty adjusting learning models to reflect user emotions, which results in low-quality training.

[0954] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0955] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on user feedback, means for certifying the AI ​​as a professional when training is complete and the goal is achieved, and means for collecting emotional data using an emotion engine that recognizes user emotions and optimizing the learning model and feedback response based on the emotional data. This enables efficient training and adjustment of high-quality professional AI specialized in a specific field by comprehensively utilizing user emotional feedback.

[0956] "General artificial intelligence" refers to an artificial intelligence system that is not limited to a specific area of ​​expertise and can perform a wide range of tasks.

[0957] "Specialized education platform" refers to a system that provides education and training to adapt general artificial intelligence to specific specialized fields.

[0958] "Industry Expert" refers to a person or group with advanced knowledge and experience in a particular area of ​​expertise.

[0959] "Training data" refers to a dataset used to learn and train artificial intelligence.

[0960] "Integrity check" refers to the process of checking the accuracy, consistency, and correctness of uploaded training data.

[0961] "Storage" refers to storing uploaded training data in a database or storage.

[0962] "Training objectives" refer to specific performance and functional indicators that an AI system must achieve.

[0963] "Standards" refers to the standards and guidelines applied in the training process.

[0964] A "learning model" refers to the structure of algorithms and parameters created by artificial intelligence using training data to perform specific tasks.

[0965] "Initialization" refers to the process of setting the parameters of a learning model to an initial state.

[0966] "Repeated learning sessions" refers to repeating the learning process multiple times using training data.

[0967] "Performance evaluation" refers to the process of measuring the performance and accuracy of an artificial intelligence system and analyzing the results.

[0968] "User" refers to the person who operates the professional education platform, provides training data, and sets training goals.

[0969] "Feedback" refers to the evaluation and suggestions provided by the user regarding the training results.

[0970] An "emotion engine" refers to a system that recognizes and analyzes emotions from a user's facial expressions, voice, and input content.

[0971] "Emotion data" refers to the user's emotion information detected by the emotion engine.

[0972] "Optimization" refers to the process of improving the performance of a learning model or system based on emotional data and feedback.

[0973] "Professional AI" refers to an artificial intelligence system that has advanced knowledge and skills in a specific specialized field and is useful in practical work.

[0974] "Certification" refers to the process of officially recognizing that training objectives have been achieved.

[0975] The specialized education platform of the present invention, which adapts a general-purpose artificial intelligence to a specific specialized field, is composed of three elements: a server, a terminal, and a user. The core of this system is the server, which receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine. Specific embodiments are described in detail below.

[0976] Overall structure

[0977] Server: The central component of the system, responsible for receiving, storing, and processing data, generating and training learning models, and performing all processes, including checking the integrity of training data, setting training goals, generating learning models, conducting repeated training sessions, evaluating performance, incorporating user feedback, and aggregating and utilizing emotional data.

[0978] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine. Users upload training data and input settings through the terminal.

[0979] User: Prepares training data in the field of expertise, sets training goals and standards, checks training status and results, provides feedback, and provides emotional feedback through the emotion engine.

[0980] Hardware / Software Used

[0981] Database: A database system for storing learning data and the training goals and criteria that are set.

[0982] Emotion engine: A software module that analyzes the user's facial expressions, voice, and input content to detect emotions.

[0983] Dashboard: A graphical user interface (GUI) for displaying the training process and the results of evaluating the AI's performance in real time.

[0984] Specific examples

[0985] Developing AI for medical translation

[0986] As an example, an embodiment in which a publisher develops a medical translation AI will be described.

[0987] 1. Data upload

[0988] User: Selects relevant materials such as medical books, papers, and medical records and uploads them to the server via the terminal.

[0989] Terminal: Provides an interface for selecting files and showing the progress of the upload.

[0990] Server: Receives the uploaded training data, checks its integrity, and stores it in the database.

[0991] 2. Setting training goals and standards

[0992] Users: Aiming to improve the accuracy of medical translations, they set standards such as consistency in translation style and terminology.

[0993] Terminal: Provides an interface for setting input and sends user input to the server.

[0994] Server: Receives the set training goals and criteria, stores them in a database, and then creates a training plan.

[0995] 3. Generating and initializing the learning model

[0996] Server: Generates and initializes the medical translation AI learning model based on medical-related data and standards, using algorithms such as the Transformer model and LSTM.

[0997] 4. Running the training process

[0998] Server: Using the training data, it runs iterative training sessions to train the AI ​​on the task of translating medical documents and optimize the model parameters.

[0999] 5. Processing emotional feedback

[1000] Device: Analyzes the user's facial expressions and voice in real time during training and collects emotional data through an emotion engine.

[1001] Server: Receives emotion data and uses it to adjust the learning model.

[1002] 6. Ratings and Feedback

[1003] Server: Evaluates the AI's performance after each training session and notifies the user of the results.

[1004] User: Modify the training data and configuration criteria as needed and submit the new configuration to the server.

[1005] 7. Certification and Deployment

[1006] Server: Once training objectives are met, certify the AI ​​as a professional and prepare it for deployment in a production environment.

[1007] Terminal: Provides an interface that notifies the user of certified information and assists in deployment in the production environment.

[1008] Prompt Sentence Examples

[1009] "Upload medical books and medical papers to start the training process of the AI ​​learning model. The training goal is to improve the accuracy of translations in the medical field."

[1010] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[1011] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1012] Step 1: Upload and manage your data

[1013] Input: Training data related to the specialty (e.g., medical books, papers, medical records).

[1014] User action: The user prepares a pre-selected training data file. For example, the user saves medical books or papers on their device.

[1015] Device operation: The user operates the device to upload the training data file to the server. The device displays a file selection screen and shows the upload progress of the file selected by the user.

[1016] Server operation: The server receives the uploaded file and checks the data for consistency (no duplicates or format inconsistencies). Once the validation is complete, the data is stored in the database.

[1017] Output: The validated training data is saved in the database.

[1018] Step 2: Setting training goals and standards

[1019] Input: Training goals and criteria (e.g., translation accuracy, style criteria).

[1020] User action: The user inputs the AI ​​training goal (e.g., "Improve the accuracy of medical translation") and training criteria (e.g., "Comply with the latest medical terminology dictionary") through the terminal.

[1021] Terminal Operation: The terminal provides an interface for inputting training goals and criteria and transmits the user's input to the server.

[1022] Server Operation: The server receives the input training goals and criteria and stores them in a database.

[1023] Output: The saved training goals and criteria are present in the database.

[1024] Step 3: Generate and initialize the learning model

[1025] Input: Training data and training target,criteria.

[1026] Server operation: The server retrieves training data, training goals, and criteria from the database, and generates an AI learning model based on them. The algorithms used may be Transformer or LSTM. The server initializes the model and sets its parameters.

[1027] Output: A trained model with initial setup completed.

[1028] Step 4: Running and monitoring the training process

[1029] Input: Initialized learning model, training data.

[1030] Server operation: The server inputs training data into the learning model and runs iterative training sessions. In each session, the AI ​​performs a specified task (e.g., translating medical documents) and optimizes the model's parameters.

[1031] Terminal operation: The terminal displays a dashboard for monitoring the training progress in real time. Users can check the progress through the terminal.

[1032] Server behavior: After a training session, the server evaluates the model's performance and records the results.

[1033] Output: Performance evaluation results and recorded data.

[1034] Step 5: Processing emotional feedback

[1035] Input: Emotional feedback (facial expressions, voice, input).

[1036] Device operation: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions.

[1037] Device operation: The detected emotion data is sent to the server.

[1038] Server operation: The server receives the user's emotional data and uses it to readjust the learning model and optimize the feedback response. For example, if the user is in a high-stress state, it will adjust the training pace.

[1039] Output: The adjusted learning model.

[1040] Step 6: Evaluate and incorporate feedback

[1041] Input: Performance evaluation results, feedback.

[1042] Server behavior: Evaluate the AI's performance after each training session. The results are displayed on a dashboard and notified to the user.

[1043] User action: The user checks the evaluation results, modifies the training data and configuration criteria as needed, and submits the new configuration to the server.

[1044] Server action: Retune the learning model based on feedback.

[1045] Output: An improved learning model.

[1046] Step 7: Certify and deploy professional AI

[1047] Input: Training goal achievement status.

[1048] Server operation: If the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI. The server exports the certified AI model in an appropriate format (e.g., API or software module) and prepares it for deployment in a production environment.

[1049] Terminal operation: Informs the user of certification information and provides an interface to assist with the steps to deploy AI in a production environment.

[1050] Output: Professional AI deployed in a production environment.

[1051] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[1052] (Application example 2)

[1053] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[1054] Currently, there are educational platforms for adapting AGI to specific specialized fields, but few systems incorporate emotional feedback, limiting their effectiveness and efficiency. In particular, in practical environments such as logistics centers, workers' emotions often have a significant impact on work efficiency. Therefore, there is a need to develop a system that combines AGI training with workers' emotional feedback to optimize work flows and improve the accuracy of learning models.

[1055] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1056] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general-purpose AI based on the set training goals and standards, means for performing repetitive learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning sessions and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is complete and the goals are achieved, and means for analyzing worker emotional feedback in real time and using the results to adjust the learning model and optimize work processes. This enables real-time work adjustments in accordance with worker emotions at logistics centers, thereby improving the accuracy of the general-purpose AI and significantly improving work efficiency.

[1057] Definition of Terms

[1058] "General artificial intelligence" is an artificial intelligence system that can adapt to a wide range of tasks and fields.

[1059] A "specialized education platform" is an educational system for providing specialized training in a specific field of expertise.

[1060] "Emotional feedback" refers to feedback data based on a user's emotional state.

[1061] An "emotion engine" is software or hardware that analyzes a user's facial expressions and voice to detect emotions.

[1062] A "logistics center" is a facility that stores, sorts, and distributes goods.

[1063] "Training data" is a data set used to train an artificial intelligence learning model.

[1064] A "learning model" is an algorithmic structure that is generated using training data to perform a specific task.

[1065] "Real-time" refers to user operations and data processing being carried out immediately without delay.

[1066] "Business process optimization" means improving processes to maximize the efficiency and effectiveness of business operations.

[1067] A "repeated learning session" is the process of repeatedly performing the same or similar tasks to train a learning model.

[1068] A "dashboard" is a graphical interface that visually displays the status and performance of a system.

[1069] "Feedback" refers to evaluations and opinions on the performance and results of a system.

[1070] "Professional certification" means officially recognizing someone as having particular skills or knowledge.

[1071] MODE FOR CARRYING OUT THE INVENTION

[1072] To implement this invention, the following system configuration is required: The system is made up of three elements: a server, a terminal, and a user. The roles and functions of each will be described in detail below.

[1073] Components

[1074] 1. Server:

[1075] Function: The server is the core of the system, receiving, storing, and processing data, generating and training learning models, evaluating them, and processing the emotion engine.

[1076] Software used: Database management software, Python

[1077] For example: The server receives training data uploaded by industry experts and stores it in a database. It then generates an AI learning model based on training goals and criteria and runs repeated learning sessions. It evaluates the performance of the learning model and sends the results to the device. It also receives emotional data sent from the device and adjusts the learning model and business process based on that data.

[1078] 2. Terminal:

[1079] Function: Provides an interface for users to operate the system, and has emotion recognition functionality using an emotion engine.

[1080] Hardware used: Smart glasses

[1081] Software used: Emotion engine, dashboard viewer

[1082] Example: A worker wears smart glasses, and the emotion engine analyzes his facial expressions and voice in real time while he works. The user sets training goals and standards through the device, uploads data, and enters feedback. In addition, the evaluation results of the learning model are displayed on a dashboard, allowing the user to check the current training progress.

[1083] 3. User:

[1084] Functions: Industry experts, providing training data and standards, emotional feedback through emotion engine.

[1085] Example: As a work manager at a logistics center, the user uploads work-related logs and data to the server, sets training goals and standards for the AI, and sends feedback and optimization of work flows to the server based on emotional data acquired during work.

[1086] Add examples to your description

[1087] Examples:

[1088] Consider a logistics center where a worker is wearing smart glasses and working in a picking area. For example, if the emotion engine detects fatigue while the worker is heading to the next picking area, it sends the result to the server. The server immediately reduces the worker's tasks temporarily and automatically assigns the tasks to a robot in the picking area. In this way, it is possible to maintain work efficiency while reducing the burden on the worker.

[1089] Example prompt sentence:

[1090] "Generate the behavior of a smart logistics system based on the following logistics center scenario. In the logistics center, workers wear smart glasses and work collaboratively with robots. Describe in detail the behavior of the system that efficiently assigns tasks based on the workers' emotional feedback and supports the work with robots."

[1091] This will enable logistics centers to adjust operations in real time according to the emotions of workers, improving the accuracy of general-purpose artificial intelligence and significantly improving operational efficiency.

[1092] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1093] Program processing steps

[1094] Step 1:

[1095] The server receives training data selected by industry experts. Users upload the training data through their devices and store it in the database after checking the integrity of the entered data. Data integrity checking includes checking the data format and error handling.

[1096] Input: Training data (e.g., logistics-related data)

[1097] Output: Consistent training data stored in a database

[1098] Step 2:

[1099] Users input training goals and standards based on their field of expertise on their terminal and send them to the server, which then sets the goals and standards that the training model must achieve.

[1100] Input: Training goals and standards (e.g., improving picking efficiency)

[1101] Output: Training goals and criteria stored in a database

[1102] Step 3:

[1103] The server generates and initializes a general AI learning model based on the uploaded training data and the set training goals and criteria. In particular, a learning model specialized for logistics operations is generated here.

[1104] Input: Stored training data, training goals and criteria

[1105] Output: Initialized training model

[1106] Step 4:

[1107] The server runs repeated training sessions on the learning model, having the model perform business tasks based on the training data, and evaluates its performance. After each session, the evaluation results are recorded in a database and sent to the device.

[1108] Input: Learning model, training data

[1109] Output: Evaluation results, updated learning model

[1110] Step 5:

[1111] The device analyzes the user's facial expressions and voice in real time through the emotion engine, generating emotion data, which is then sent to the server.

[1112] Input: User's facial expression, voice

[1113] Output: Emotion data

[1114] Step 6:

[1115] The server receives the acquired emotion data and reflects it in adjusting the performance of the learning model and the workflow. Specifically, it takes operational optimization measures such as reallocating tasks according to the worker's emotions.

[1116] Input: Emotion data, learning model

[1117] Output: Optimized workflow, adjusted learning model

[1118] Step 7:

[1119] Once training is complete and the objectives are met, the server executes the process of certifying the AI ​​as a professional, and the certified AI is deployed in a practical environment such as a logistics center.

[1120] Input: Results of completed training sessions, evaluation data

[1121] Output: Certified artificial intelligence, ready for production deployment

[1122] This allows the server, terminal, and user to work together to realize learning and optimization of general-purpose artificial intelligence based on emotional feedback, enabling efficient business operations at logistics centers.

[1123] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1124] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1125] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1126] [Third embodiment]

[1127] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1128] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1129] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1130] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1131] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1132] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1133] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1134] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1135] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1136] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1137] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1138] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1139] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1140] Overall system configuration

[1141] Server: The core of the system, it receives, stores, processes data, generates, trains, and evaluates learning models. The server manages multiple AI learning models and serves as a specialized education platform.

[1142] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc.

[1143] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[1144] Program processing

[1145] 1. Data upload and management

[1146] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[1147] Terminal: The user uploads the selected training data file to the server via the terminal.

[1148] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[1149] 2. Setting training goals and standards

[1150] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[1151] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[1152] Server: Receives training goals and criteria and stores them in a database.

[1153] 3. Generating and initializing the learning model

[1154] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the generated learning model and performs initial settings (e.g., setting initial parameter values, setting hyperparameters for training).

[1155] 4. Running and monitoring the training process

[1156] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[1157] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[1158] User: View the AI's performance during training on a dashboard and provide feedback as needed. If the training data or criteria need to be adjusted, send new instructions to the server.

[1159] 5. Evaluation and feedback

[1160] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[1161] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1162] 6. Certification and deployment of professional AI

[1163] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1164] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1165] Specific examples

[1166] For example, if a publisher develops a medical translation AI, the following embodiment applies:

[1167] 1. Data upload:

[1168] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[1169] 2. Setting training goals and standards:

[1170] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[1171] 3. Generating and initializing the learning model:

[1172] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[1173] 4. Running the training process:

[1174] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1175] 5. Evaluation and feedback:

[1176] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[1177] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[1178] 6. Certification and deployment of professional AI:

[1179] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[1180] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1181] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[1182] The processing flow will be explained below.

[1183] Step 1:

[1184] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[1185] Step 2:

[1186] Device: The user uploads the selected training data file to the server via the device. An upload UI is provided to assist with file selection and transmission.

[1187] Step 3:

[1188] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the data and automatically inspects the format and content of the data before storing it.

[1189] Step 4:

[1190] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[1191] Step 5:

[1192] Terminal: Provides a UI for inputting training goals and criteria, sends the settings to the server, and displays a confirmation message to the user when the sending operation is complete.

[1193] Step 6:

[1194] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[1195] Step 7:

[1196] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[1197] Step 8:

[1198] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[1199] Step 9:

[1200] Server: Automatically evaluates the AI's performance after each training session, records the results, scores them based on the evaluation criteria, and stores them in a database.

[1201] Step 10:

[1202] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[1203] Step 11:

[1204] User: Check the AI's performance during training on the dashboard and provide feedback as needed. Feedback is sent from the device to the server.

[1205] Step 12:

[1206] Server: After each training session, the server evaluates the AI's performance and uses the feedback to retune the learning model, resetting parameters as needed.

[1207] Step 13:

[1208] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1209] Step 14:

[1210] Server: Receives new settings from the user, adjusts the training plan again, and continues the learning session.

[1211] Step 15:

[1212] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1213] Step 16:

[1214] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​works effectively in the field.

[1215] Example 1

[1216] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1217] Adapting general-purpose AI to specific specialized fields requires effective management of diverse training data and the generation of highly accurate learning models. However, conventional systems have not adequately verified the consistency of specialized training data, monitored the training process in real time, or applied the data to a practical environment after the training goal has been achieved. As a result, it has been difficult to efficiently and effectively develop specialized AI and deploy it in practice.

[1218] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1219] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model of the general purpose AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is completed and the goal is achieved, and means for exporting the generated professional AI model for application to a practical environment. This makes it possible to efficiently and effectively adapt the general purpose AI to a specific field of expertise and to develop and deploy professional AI that is suited to practical use.

[1220] An "industry expert" is someone who has extensive knowledge and experience in a particular field and is able to select training data and set standards in that field.

[1221] "Training data" is a collection of data used to train an artificial intelligence to acquire the knowledge necessary to perform a specific task.

[1222] "Integrity check" refers to the process of verifying that uploaded data is in the correct format, free of missing or duplicated data, and suitable for training.

[1223] "Storage" refers to the technology used to hold uploaded data using a database or storage system.

[1224] "Training goals" refer to the specific performance or functional goals that artificial intelligence should achieve through training.

[1225] "Training standards" refer to the guidelines and rules necessary to achieve training goals, including standards for translation style and accuracy.

[1226] A "learning model" refers to an algorithm or neural network that artificial intelligence generates based on training data.

[1227] "Initialization" refers to the process of initializing the generated learning model and preparing it for training.

[1228] "Iterative learning sessions" refers to the learning process of using training data to update the learning model over multiple cycles to improve performance.

[1229] "Performance evaluation" refers to the process of calculating indicators to measure how close a learning model is to achieving its goals and analyzing the results.

[1230] "Means for notifying users" refers to the technology and methods for notifying users of performance evaluation results via a dashboard or notification function.

[1231] "Adjusting based on feedback" refers to the process of changing the parameters and training data of a learning model based on user opinions and evaluations to improve the accuracy of the model.

[1232] "Professional certification" refers to the process of officially recognizing artificial intelligence that has met its training objectives as suitable for use in practice in a particular professional field.

[1233] "Export means" refers to the technology or method for converting and outputting a certified AI model into an appropriate data format for application in a production environment.

[1234] "Applying to a production environment" refers to the process of actually implementing and using a certified AI model in a specific business or task.

[1235] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1236] Overall system configuration

[1237] Server: This is the core of the system and has the functions of receiving, storing, and processing data, as well as generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Suitable software for use includes MySQL for database management and TensorFlow or PyTorch for generating and training AI models.

[1238] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc. The interface is generally provided through a web browser.

[1239] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[1240] Program processing

[1241] The server processes the program as follows:

[1242] Data Upload and Management:

[1243] Users prepare training data related to their field of expertise. For example, if training a medical translation AI, they would select medical books, papers, medical records, and terminology dictionaries.

[1244] Through the device's web browser, the user uploads the selected training data file, for example, using an HTML5-based file upload widget.

[1245] The server receives uploaded training data, stores it in a MySQL database, and uses Python scripts to check the data for consistency and correctness.

[1246] Setting training goals and standards:

[1247] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI.

[1248] The terminal provides a form for transmitting the entered training goals and criteria to the server.

[1249] The server validates the received training goals and criteria and stores them in a database.

[1250] Generate and initialize the learning model:

[1251] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[1252] Running and monitoring the training process:

[1253] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, a Python script automatically evaluates the AI's performance and records the results.

[1254] The terminal provides a dashboard for monitoring the training process in real time and displays the evaluation results using JavaScript.

[1255] Users can check the AI's performance during training on a dashboard and provide feedback as needed. If necessary, they can input training data or instructions for adjusting the criteria into the device and send them to the server.

[1256] Rating and feedback:

[1257] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[1258] Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server.

[1259] Certification and deployment of professional AI:

[1260] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[1261] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[1262] Specific examples

[1263] As an example, we will explain a specific embodiment in which a publisher develops a medical translation AI.

[1264] 1. Data upload:

[1265] User: Selects learning data such as medical books, papers, and medical records and uploads them to the server via the device.

[1266] 2. Setting training goals and standards:

[1267] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[1268] 3. Generating and initializing the learning model:

[1269] Server: Generates and initializes AI learning models using TensorFlow and PyTorch based on medical-related data and standards.

[1270] 4. Running the training process:

[1271] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1272] 5. Evaluation and feedback:

[1273] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[1274] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[1275] 6. Certification and deployment of professional AI:

[1276] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[1277] On the device: Inform the user of the authentication information and provide an expand button.

[1278] Example prompts to input to the generative AI model

[1279] Please generate a learning model specialized for medical translation AI. Using the following data, the training goal is "improving translation accuracy" and the evaluation criteria are "accuracy of terminology and matching of context." Upload data includes medical books, papers, and medical records. Please also perform the initial setup of the training data.

[1280] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[1281] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1282] Step 1: Data upload and management

[1283] The user prepares learning data related to their field of expertise (e.g., medical books, papers, medical records, glossaries). Next, they open a web browser on their device, access the system's data upload screen, click the "Select File" button, and select the learning data file they have selected. Then, they press the "Upload" button to send the data to the server.

[1284] Input: Training data files such as medical books, papers, and medical records

[1285] Data processing and calculation: Check file format, detect missing data, delete duplicate data

[1286] Output: The training data with consistency confirmed is saved in the database.

[1287] Specific operation: The server stores the uploaded file in temporary storage, runs a Python script to check the file format, check for missing data, and remove duplicates, and then stores it in the official database.

[1288] Step 2: Setting training goals and standards

[1289] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI. Users enter the training goals and criteria in the text fields and execute the settings by pressing the save button.

[1290] Input: Training goals and standards (e.g., "Achieve medical translation accuracy of 95% or higher," "Use technical terms preferentially")

[1291] Data processing and data calculation: Validation of input data (format confirmation and content consistency confirmation)

[1292] Output: Training goals and criteria are stored in a database

[1293] Specific behavior: The server checks the received training goals and criteria with validation scripts (e.g., checking for invalid input), executes SQL queries, and records them in the database.

[1294] Step 3: Generate and initialize the learning model

[1295] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[1296] Input: Consistent training data, training criteria

[1297] Data processing and data calculation: Defining model architecture, initializing parameters, setting hyperparameters

[1298] Output: Initialized AI learning model

[1299] What happens: The server runs a Python script to define the layers of the neural network and set the initial parameters, so the learning model is ready to start training.

[1300] Step 4: Running and monitoring the training process

[1301] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, the AI's performance is automatically evaluated and the results are recorded.

[1302] Input: Consistency-checked training data, initialized AI learning model

[1303] Data processing and calculation: data batching, error calculation, backpropagation, model updating

[1304] Output: Performance evaluation results after each session

[1305] Specific operation: The server divides the training data into batches and inputs them into the model, calculates the error function for each batch, performs backpropagation to update the model, and records the training progress and evaluation results in a log file.

[1306] Step 5: Evaluate and incorporate feedback

[1307] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[1308] Input: performance results after training sessions, user feedback

[1309] Data processing and data calculation: Calculation of evaluation metrics (e.g., precision, recall, F-measure), analysis of feedback, model retraining

[1310] Output: Training results displayed in a dashboard, with an updated model incorporating feedback

[1311] Specific operation: The server executes the evaluation script to evaluate the model's performance, records the results in a database, and displays them on a dashboard. It also performs retraining and parameter adjustment based on user feedback.

[1312] Step 6: Certification and deployment of professional AI

[1313] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[1314] Input: Completely trained AI model, training result data

[1315] Data processing and calculation: Final adjustment of the model, export of the model

[1316] Output: AI models exported in a format suitable for production environments

[1317] Specific operation: The server runs an auxiliary script to evaluate the training results, and if it determines that the goal has been achieved, it exports the model in ONNX format and packages it into a Docker container.

[1318] Step 7: Notification of certification information and deployment of AI

[1319] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[1320] Input: Certified AI model, user deployment instructions

[1321] Data processing and data calculation: Notification of certification information, settings for deployment

[1322] Output: AI model deployed in production

[1323] Specific operation: The user checks the notification on the device and clicks the deploy button to deploy the AI ​​model to the production environment. The server receives the instruction and actually performs the deployment.

[1324] (Application example 1)

[1325] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1326] Conventional methods for training and evaluating AI robots specialized for specific tasks in factories are time-consuming and costly, preventing efficient automation. Furthermore, it is difficult to monitor the training process and check evaluation results in real time, making it difficult to provide appropriate feedback in a timely manner. A method for solving these problems and developing AI robots specialized in specific fields is needed.

[1327] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1328] In this invention, the server includes: means for uploading training data selected by industry experts; means for verifying the integrity of the uploaded training data and storing it; means for setting training goals and standards based on the field of expertise; means for generating and initializing a learning model for a general AI based on the set training goals and standards; means for performing repeated learning sessions on the learning model using the training data; means for evaluating the performance of the AI ​​after the learning sessions and notifying a user of the results; means for adjusting the learning model based on user feedback; means for certifying the AI ​​as a professional when training is complete and the goals are achieved; means for training an AI robot specialized in a specific task to automate work in a factory; and means for setting training data and training goals using a smartphone and checking the training process and evaluation results in real time, which not only significantly improves work efficiency in the factory and reduces time and costs, but also enables real-time monitoring of the training process and checking the evaluation results.

[1329] "General purpose artificial intelligence" refers to artificial intelligence that has a wide range of applications, not limited to specific tasks or fields.

[1330] A "specialized education platform" is a system for training and educating artificial intelligence specialized in specific fields.

[1331] An "industry expert" is an expert with advanced knowledge and experience in a particular field of expertise.

[1332] "Training data" refers to the data set used to train an artificial intelligence.

[1333] "Upload" refers to the act of transferring data from a local terminal to a server.

[1334] "Integrity" means that data is correct, consistent, and free from errors.

[1335] "Preservation" refers to the act of storing data for a long period of time.

[1336] A "training goal" is a specific goal that you want to achieve in learning artificial intelligence.

[1337] "Training standards" refer to the standards and rules that artificial intelligence must follow during the training process.

[1338] A "learning model" is an artificial intelligence algorithm built based on training data.

[1339] "Initialization" refers to the act of resetting the parameters of a learning model to their initial settings.

[1340] A "learning session" is the process by which an artificial intelligence repeatedly learns using training data.

[1341] "Performance evaluation" refers to evaluating the performance of a trained artificial intelligence.

[1342] "User" refers to a person or institution using the system.

[1343] "Feedback" refers to evaluations and comments on training results provided by users.

[1344] "Accreditation" is the act of officially recognizing a qualification or status based on specific criteria.

[1345] A "professional" is a person who has advanced knowledge and skills in a specific specialized field and is engaged in that occupation, or a system that supports such a person.

[1346] "Factory work" refers to a series of tasks and processes carried out on the factory production floor.

[1347] "Automation" is the act of enabling machines or systems to perform tasks autonomously without human intervention.

[1348] A "smartphone" is a mobile phone that has computing capabilities.

[1349] "Real time" means that events are processed and displayed as they occur.

[1350] A "dashboard" is a user interface that visually displays the status and data of a system.

[1351] The professional education platform of the present invention is designed for the purpose of training artificial intelligence robots to automate specific tasks in factories. This system mainly consists of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1352] Overall system configuration

[1353] server

[1354] The server is the core of this system and has the functions of receiving, storing, and processing data, generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Specifically, the server uses the following hardware and software:

[1355] Hardware: High performance computer, GPU (e.g. NVIDIA GPU)

[1356] Software: Python, Flask, PyTorch, Transformers library

[1357] Terminal

[1358] The terminal provides an interface for users to operate the system. Smartphones are mainly used. The terminal provides the following functions:

[1359] Uploading data

[1360] Setting training goals and standards

[1361] Monitoring the training process

[1362] Providing Feedback

[1363] User

[1364] Users are experts in factory operations and provide training data and standards, as well as feedback to the AI ​​during training.

[1365] Program processing

[1366] Data Upload and Management

[1367] User: The user prepares training data related to factory operations, such as product inspection images and sensor data.

[1368] Terminal: Upload the selected training data file to the server via the terminal.

[1369] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[1370] Setting training goals and standards

[1371] User: The user inputs the AI ​​training goal (e.g., improving the accuracy of product inspection) and training criteria (e.g., inspection speed, false positive rate) from the terminal.

[1372] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[1373] Server: Receives training goals and criteria and stores them in a database.

[1374] Generating and initializing the learning model

[1375] Server: Generates a general-purpose AI learning model based on the training data and standards received from the user. Initializes the generated learning model and performs initial settings.

[1376] Running and monitoring the training process

[1377] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., product inspection) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[1378] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[1379] Evaluation and feedback

[1380] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[1381] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1382] Certification and deployment of professional AI

[1383] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1384] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1385] Specific examples

[1386] As an example, consider an AI robot that automates the task of inspecting products on a conveyor belt in a factory. First, the user uploads the image data of the product inspection to a server using a smartphone. Next, the user inputs training goals and criteria for the AI ​​to achieve an accuracy of 95% or more in product inspection. Once the training process is complete, the user can view the evaluation results on a dashboard and provide feedback.

[1387] Example prompt sentence:

[1388] "Start the Factory Robot AI Trainer app and upload the image data of the object to be inspected. Then, enter the training goals and criteria for the AI ​​robot to achieve a product inspection accuracy of 95% or more. After training is complete, review the evaluation results and provide feedback."

[1389] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1390] Step 1:

[1391] Data upload

[1392] Users prepare training data related to factory operations (e.g., product inspection images and sensor data) and upload it to the server via their terminal. The terminal checks the file format and size of the uploaded data and transfers it to the server. The server receives the file and stores it in a database. The input is the training data, and the output is the data stored in the server's database.

[1393] Step 2:

[1394] Setting training goals and standards

[1395] The user inputs training goals (e.g., improving product inspection accuracy) and training criteria (e.g., inspection speed, false positive rate) from the terminal. The terminal sends the goals and criteria entered by the user to the server. The server receives this data and stores it in a database. The input is the goals and criteria entered by the user, and the output is the training goals and criteria stored in the server's database.

[1396] Step 3:

[1397] Generating and initializing the learning model

[1398] The server generates and initializes a general-purpose AI learning model based on the uploaded training data and the set goals and criteria. Specifically, it uses Python, PyTorch, and the Transformers library to build the learning model and set initial parameters. The input is the training data, goals, and criteria, and the output is the initialized learning model.

[1399] Step 4:

[1400] Running and monitoring the learning process

[1401] The server runs repeated learning sessions on the generated learning model. Specifically, the AI ​​performs a specific task (e.g., product inspection) using the uploaded data and records the results. After each session, the server evaluates the AI's performance and records the results. The device monitors this process in real time and displays the evaluation results on a user interface (dashboard). The inputs are the training data and the learning model, and the output is the evaluation results.

[1402] Step 5:

[1403] Evaluation and feedback

[1404] The server evaluates the AI's performance after each training session and displays the evaluation results on a dashboard. The user can check the evaluation results on the dashboard and provide feedback as needed. The server receives feedback from the user and readjusts the training model. The inputs are the evaluation results and user feedback, and the output is the adjusted training model.

[1405] Step 6:

[1406] Certification and deployment of professional AI

[1407] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI. The certified AI model is exported in an appropriate format and prepared for deployment in a production environment. The terminal notifies the user of the certification information and deploys the AI ​​in a production environment as necessary. The input is the achievement status of the training goals, and the output is the certified AI model and notification.

[1408] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1409] The specialized training platform for adapting general-purpose AI to specific specialized fields according to the present invention is composed of three elements: a server, a terminal, and a user. Furthermore, the system combines an emotion engine that recognizes user emotions and utilizes user emotional feedback in the AI ​​training process.

[1410] Overall system configuration

[1411] Server: The core of the system, it receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine.

[1412] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine.

[1413] Users: Industry experts who provide training data and standards, and emotional feedback from the emotion engine.

[1414] Emotion engine: Analyzes the user's facial expressions, voice, and input to detect emotions and reflect them in performance and feedback.

[1415] Program processing

[1416] 1. Data upload and management

[1417] User: Prepare training data related to their field of expertise. For example, if training a medical translation AI, select medical books, papers, medical records, and terminology dictionaries.

[1418] Terminal: The user uploads the selected training data file to the server through the terminal. The terminal interface assists the upload operation.

[1419] Server: Receives uploaded training data and stores it in the database. Checks the integrity and correctness of the data and stores it in the database.

[1420] 2. Setting training goals and standards

[1421] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context) from the device.

[1422] Terminal: Provides an interface for inputting training goals and criteria and transmits the settings to the server.

[1423] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[1424] 3. Generating and initializing the learning model

[1425] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[1426] 4. Running and monitoring the training process

[1427] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data. After each training session, the AI's performance is automatically evaluated and the results are recorded.

[1428] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[1429] 5. Processing emotional feedback

[1430] Device: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The emotion recognition results are sent to the server.

[1431] Server: Receives user emotion data and uses it to adjust learning models and optimize feedback responses.

[1432] 6. Evaluation and feedback

[1433] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. Readjust the learning model based on the user's emotional data.

[1434] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1435] 7. Certification and deployment of professional AI

[1436] Server: Once the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1437] Terminal: Informs users of certification information and assists them in deploying AI in production environments as needed.

[1438] Specific examples

[1439] For example, if a publisher develops a medical translation AI, the following embodiment applies.

[1440] 1. Data upload:

[1441] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[1442] 2. Setting training goals and standards:

[1443] User: Enter the goals for improving the accuracy of AI medical translation and translation style standards on the device, and send the settings to the server.

[1444] 3. Generating and initializing the learning model:

[1445] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[1446] 4. Running the training process:

[1447] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1448] 5. Processing emotional feedback:

[1449] Terminal: Recognizes the user's emotions in real time during training and sends the results to the server via the emotion engine.

[1450] Server: Leverages emotional data to adjust learning models and feedback responses.

[1451] 6. Evaluation and Feedback:

[1452] Server: Evaluates the training results and retunes the model based on user emotional feedback.

[1453] User: Modify the training data or criteria as needed and resubmit.

[1454] 7. Certification and deployment of professional AI:

[1455] Server: Executes the certification process when training objectives are achieved and deploys to the production environment.

[1456] Terminal: Notifies certification information and assists with AI deployment.

[1457] This allows the system to efficiently and effectively specialize general AI in specific fields of expertise, and comprehensively utilize user feedback to develop high-quality professional AI.

[1458] The processing flow will be explained below.

[1459] Step 1:

[1460] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[1461] Step 2:

[1462] Terminal: The user uses an interface to upload the selected training data file to the server through the terminal. The user is assisted in selecting the file and pressing the upload button.

[1463] Step 3:

[1464] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the uploaded data and inspects the format and content for any problems.

[1465] Step 4:

[1466] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[1467] Step 5:

[1468] Terminal: Provides an interface for inputting training goals and criteria, transmits the settings to the server, and displays a confirmation message to the user upon completion.

[1469] Step 6:

[1470] Server: Receives training goals and criteria and stores them in a database. Creates a training plan based on the received settings information.

[1471] Step 7:

[1472] Server: Generates and initializes a general-purpose AI learning model based on the learning data and training criteria received from the user. Initialization involves setting model parameters and loading initial data.

[1473] Step 8:

[1474] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[1475] Step 9:

[1476] Server: After each training session, the server automatically evaluates the AI's performance and records the results. The evaluation is based on set criteria and saved in the form of a score.

[1477] Step 10:

[1478] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to see training progress and performance.

[1479] Step 11:

[1480] Device: During training, the emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The detected emotion data is sent to the server.

[1481] Step 12:

[1482] Server: Receives and records the emotional data sent by users along with their training performance and feedback. The emotional data is used as a factor in adjusting the learning model.

[1483] Step 13:

[1484] User: Check the AI's performance during training on the dashboard and send feedback to the server from the device as needed. Feedback can be entered as text or multiple choice options.

[1485] Step 14:

[1486] Server: Retunes the learning model based on the evaluation results and sentiment data collected after each training session. Retunement includes parameter resetting and model retraining.

[1487] Step 15:

[1488] User: Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server via the terminal.

[1489] Step 16:

[1490] Server: Receives new settings from the user, adjusts the training plan again and continues the learning session, saves the settings in the database, and starts a new training session.

[1491] Step 17:

[1492] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1493] Step 18:

[1494] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​can function effectively in the field.

[1495] Example 2

[1496] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1497] A specialized training platform for effectively adapting general AI to specific specialized fields must efficiently manage the entire process from selecting training data, verifying data integrity, setting training goals and standards, running learning sessions, and evaluating and certifying performance, while incorporating and optimizing the platform by incorporating user emotional feedback.Current technology has difficulty adjusting learning models to reflect user emotions, which results in low-quality training.

[1498] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1499] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on user feedback, means for certifying the AI ​​as a professional when training is complete and the goal is achieved, and means for collecting emotional data using an emotion engine that recognizes user emotions and optimizing the learning model and feedback response based on the emotional data. This enables efficient training and adjustment of high-quality professional AI specialized in a specific field by comprehensively utilizing user emotional feedback.

[1500] "General artificial intelligence" refers to an artificial intelligence system that is not limited to a specific area of ​​expertise and can perform a wide range of tasks.

[1501] "Specialized education platform" refers to a system that provides education and training to adapt general artificial intelligence to specific specialized fields.

[1502] "Industry Expert" refers to a person or group with advanced knowledge and experience in a particular area of ​​expertise.

[1503] "Training data" refers to a dataset used to learn and train artificial intelligence.

[1504] "Integrity check" refers to the process of checking the accuracy, consistency, and correctness of uploaded training data.

[1505] "Storage" refers to storing uploaded training data in a database or storage.

[1506] "Training objectives" refer to specific performance and functional indicators that an AI system must achieve.

[1507] "Standards" refers to the standards and guidelines applied in the training process.

[1508] A "learning model" refers to the structure of algorithms and parameters created by artificial intelligence using training data to perform specific tasks.

[1509] "Initialization" refers to the process of setting the parameters of a learning model to an initial state.

[1510] "Repeated learning sessions" refers to repeating the learning process multiple times using training data.

[1511] "Performance evaluation" refers to the process of measuring the performance and accuracy of an artificial intelligence system and analyzing the results.

[1512] "User" refers to the person who operates the professional education platform, provides training data, and sets training goals.

[1513] "Feedback" refers to the evaluation and suggestions provided by the user regarding the training results.

[1514] An "emotion engine" refers to a system that recognizes and analyzes emotions from a user's facial expressions, voice, and input content.

[1515] "Emotion data" refers to the user's emotion information detected by the emotion engine.

[1516] "Optimization" refers to the process of improving the performance of a learning model or system based on emotional data and feedback.

[1517] "Professional AI" refers to an artificial intelligence system that has advanced knowledge and skills in a specific specialized field and is useful in practical work.

[1518] "Certification" refers to the process of officially recognizing that training objectives have been achieved.

[1519] The specialized education platform of the present invention, which adapts a general-purpose artificial intelligence to a specific specialized field, is composed of three elements: a server, a terminal, and a user. The core of this system is the server, which receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine. Specific embodiments are described in detail below.

[1520] Overall structure

[1521] Server: The central component of the system, responsible for receiving, storing, and processing data, generating and training learning models, and performing all processes, including checking the integrity of training data, setting training goals, generating learning models, conducting repeated training sessions, evaluating performance, incorporating user feedback, and aggregating and utilizing emotional data.

[1522] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine. Users upload training data and input settings through the terminal.

[1523] User: Prepares training data in the field of expertise, sets training goals and standards, checks training status and results, provides feedback, and provides emotional feedback through the emotion engine.

[1524] Hardware / Software Used

[1525] Database: A database system for storing learning data and the training goals and criteria that are set.

[1526] Emotion engine: A software module that analyzes the user's facial expressions, voice, and input content to detect emotions.

[1527] Dashboard: A graphical user interface (GUI) for displaying the training process and the results of evaluating the AI's performance in real time.

[1528] Specific examples

[1529] Developing AI for medical translation

[1530] As an example, an embodiment in which a publisher develops a medical translation AI will be described.

[1531] 1. Data upload

[1532] User: Selects relevant materials such as medical books, papers, and medical records and uploads them to the server via the terminal.

[1533] Terminal: Provides an interface for selecting files and showing the progress of the upload.

[1534] Server: Receives the uploaded training data, checks its integrity, and stores it in the database.

[1535] 2. Setting training goals and standards

[1536] Users: Aiming to improve the accuracy of medical translations, they set standards such as consistency in translation style and terminology.

[1537] Terminal: Provides an interface for setting input and sends user input to the server.

[1538] Server: Receives the set training goals and criteria, stores them in a database, and then creates a training plan.

[1539] 3. Generating and initializing the learning model

[1540] Server: Generates and initializes the medical translation AI learning model based on medical-related data and standards, using algorithms such as the Transformer model and LSTM.

[1541] 4. Running the training process

[1542] Server: Using the training data, it runs iterative training sessions to train the AI ​​on the task of translating medical documents and optimize the model parameters.

[1543] 5. Processing emotional feedback

[1544] Device: Analyzes the user's facial expressions and voice in real time during training and collects emotional data through an emotion engine.

[1545] Server: Receives emotion data and uses it to adjust the learning model.

[1546] 6. Ratings and Feedback

[1547] Server: Evaluates the AI's performance after each training session and notifies the user of the results.

[1548] User: Modify the training data and configuration criteria as needed and submit the new configuration to the server.

[1549] 7. Certification and Deployment

[1550] Server: Once training objectives are met, certify the AI ​​as a professional and prepare it for deployment in a production environment.

[1551] Terminal: Provides an interface that notifies the user of certified information and assists in deployment in the production environment.

[1552] Prompt Sentence Examples

[1553] "Upload medical books and medical papers to start the training process of the AI ​​learning model. The training goal is to improve the accuracy of translations in the medical field."

[1554] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[1555] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1556] Step 1: Upload and manage your data

[1557] Input: Training data related to the specialty (e.g., medical books, papers, medical records).

[1558] User action: The user prepares a pre-selected training data file. For example, the user saves medical books or papers on their device.

[1559] Device operation: The user operates the device to upload the training data file to the server. The device displays a file selection screen and shows the upload progress of the file selected by the user.

[1560] Server operation: The server receives the uploaded file and checks the data for consistency (no duplicates or format inconsistencies). Once the validation is complete, the data is stored in the database.

[1561] Output: The validated training data is saved in the database.

[1562] Step 2: Setting training goals and standards

[1563] Input: Training goals and criteria (e.g., translation accuracy, style criteria).

[1564] User action: The user inputs the AI ​​training goal (e.g., "Improve the accuracy of medical translation") and training criteria (e.g., "Comply with the latest medical terminology dictionary") through the terminal.

[1565] Terminal Operation: The terminal provides an interface for inputting training goals and criteria and transmits the user's input to the server.

[1566] Server Operation: The server receives the input training goals and criteria and stores them in a database.

[1567] Output: The saved training goals and criteria are present in the database.

[1568] Step 3: Generate and initialize the learning model

[1569] Input: Training data and training target,criteria.

[1570] Server operation: The server retrieves training data, training goals, and criteria from the database, and generates an AI learning model based on them. The algorithms used may be Transformer or LSTM. The server initializes the model and sets its parameters.

[1571] Output: A trained model with initial setup completed.

[1572] Step 4: Running and monitoring the training process

[1573] Input: Initialized learning model, training data.

[1574] Server operation: The server inputs training data into the learning model and runs iterative training sessions. In each session, the AI ​​performs a specified task (e.g., translating medical documents) and optimizes the model's parameters.

[1575] Terminal operation: The terminal displays a dashboard for monitoring the training progress in real time. Users can check the progress through the terminal.

[1576] Server behavior: After a training session, the server evaluates the model's performance and records the results.

[1577] Output: Performance evaluation results and recorded data.

[1578] Step 5: Processing emotional feedback

[1579] Input: Emotional feedback (facial expressions, voice, input).

[1580] Device operation: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions.

[1581] Device operation: The detected emotion data is sent to the server.

[1582] Server operation: The server receives the user's emotional data and uses it to readjust the learning model and optimize the feedback response. For example, if the user is in a high-stress state, it will adjust the training pace.

[1583] Output: The adjusted learning model.

[1584] Step 6: Evaluate and incorporate feedback

[1585] Input: Performance evaluation results, feedback.

[1586] Server behavior: Evaluate the AI's performance after each training session. The results are displayed on a dashboard and notified to the user.

[1587] User action: The user checks the evaluation results, modifies the training data and configuration criteria as needed, and submits the new configuration to the server.

[1588] Server action: Retune the learning model based on feedback.

[1589] Output: An improved learning model.

[1590] Step 7: Certify and deploy professional AI

[1591] Input: Training goal achievement status.

[1592] Server operation: If the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI. The server exports the certified AI model in an appropriate format (e.g., API or software module) and prepares it for deployment in a production environment.

[1593] Terminal operation: Informs the user of certification information and provides an interface to assist with the steps to deploy AI in a production environment.

[1594] Output: Professional AI deployed in a production environment.

[1595] This will enable comprehensive utilization of user emotional feedback to efficiently and effectively develop high-quality professional AI specialized in specific fields.

[1596] (Application example 2)

[1597] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1598] Currently, there are educational platforms for adapting AGI to specific specialized fields, but few systems incorporate emotional feedback, limiting their effectiveness and efficiency. In particular, in practical environments such as logistics centers, workers' emotions often have a significant impact on work efficiency. Therefore, there is a need to develop a system that combines AGI training with workers' emotional feedback to optimize work flows and improve the accuracy of learning models.

[1599] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1600] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model for the general-purpose AI based on the set training goals and standards, means for performing repetitive learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning sessions and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is complete and the goals are achieved, and means for analyzing worker emotional feedback in real time and using the results to adjust the learning model and optimize work processes. This enables real-time work adjustments in accordance with worker emotions at logistics centers, thereby improving the accuracy of the general-purpose AI and significantly improving work efficiency.

[1601] Definition of Terms

[1602] "General artificial intelligence" is an artificial intelligence system that can adapt to a wide range of tasks and fields.

[1603] A "specialized education platform" is an educational system for providing specialized training in a specific field of expertise.

[1604] "Emotional feedback" refers to feedback data based on a user's emotional state.

[1605] An "emotion engine" is software or hardware that analyzes a user's facial expressions and voice to detect emotions.

[1606] A "logistics center" is a facility that stores, sorts, and distributes goods.

[1607] "Training data" is a data set used to train an artificial intelligence learning model.

[1608] A "learning model" is an algorithmic structure that is generated using training data to perform a specific task.

[1609] "Real-time" refers to user operations and data processing being carried out immediately without delay.

[1610] "Business process optimization" means improving processes to maximize the efficiency and effectiveness of business operations.

[1611] A "repeated learning session" is the process of repeatedly performing the same or similar tasks to train a learning model.

[1612] A "dashboard" is a graphical interface that visually displays the status and performance of a system.

[1613] "Feedback" refers to evaluations and opinions on the performance and results of a system.

[1614] "Professional certification" means officially recognizing someone as having particular skills or knowledge.

[1615] MODE FOR CARRYING OUT THE INVENTION

[1616] To implement this invention, the following system configuration is required: The system is made up of three elements: a server, a terminal, and a user. The roles and functions of each will be described in detail below.

[1617] Components

[1618] 1. Server:

[1619] Function: The server is the core of the system, receiving, storing, and processing data, generating and training learning models, evaluating them, and processing the emotion engine.

[1620] Software used: Database management software, Python

[1621] For example: The server receives training data uploaded by industry experts and stores it in a database. It then generates an AI learning model based on training goals and criteria and runs repeated learning sessions. It evaluates the performance of the learning model and sends the results to the device. It also receives emotional data sent from the device and adjusts the learning model and business process based on that data.

[1622] 2. Terminal:

[1623] Function: Provides an interface for users to operate the system, and has emotion recognition functionality using an emotion engine.

[1624] Hardware used: Smart glasses

[1625] Software used: Emotion engine, dashboard viewer

[1626] Example: A worker wears smart glasses, and the emotion engine analyzes his facial expressions and voice in real time while he works. The user sets training goals and standards through the device, uploads data, and enters feedback. In addition, the evaluation results of the learning model are displayed on a dashboard, allowing the user to check the current training progress.

[1627] 3. User:

[1628] Functions: Industry experts, providing training data and standards, emotional feedback through emotion engine.

[1629] Example: As a work manager at a logistics center, the user uploads work-related logs and data to the server, sets training goals and standards for the AI, and sends feedback and optimization of work flows to the server based on emotional data acquired during work.

[1630] Add examples to your description

[1631] Examples:

[1632] Consider a logistics center where a worker is wearing smart glasses and working in a picking area. For example, if the emotion engine detects fatigue while the worker is heading to the next picking area, it sends the result to the server. The server immediately reduces the worker's tasks temporarily and automatically assigns the tasks to a robot in the picking area. In this way, it is possible to maintain work efficiency while reducing the burden on the worker.

[1633] Example prompt sentence:

[1634] "Generate the behavior of a smart logistics system based on the following logistics center scenario. In the logistics center, workers wear smart glasses and work collaboratively with robots. Describe in detail the behavior of the system that efficiently assigns tasks based on the workers' emotional feedback and supports the work with robots."

[1635] This will enable logistics centers to adjust operations in real time according to the emotions of workers, improving the accuracy of general-purpose artificial intelligence and significantly improving operational efficiency.

[1636] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1637] Program processing steps

[1638] Step 1:

[1639] The server receives training data selected by industry experts. Users upload the training data through their devices and store it in the database after checking the integrity of the entered data. Data integrity checking includes checking the data format and error handling.

[1640] Input: Training data (e.g., logistics-related data)

[1641] Output: Consistent training data stored in a database

[1642] Step 2:

[1643] Users input training goals and standards based on their field of expertise on their terminal and send them to the server, which then sets the goals and standards that the training model must achieve.

[1644] Input: Training goals and standards (e.g., improving picking efficiency)

[1645] Output: Training goals and criteria stored in a database

[1646] Step 3:

[1647] The server generates and initializes a general AI learning model based on the uploaded training data and the set training goals and criteria. In particular, a learning model specialized for logistics operations is generated here.

[1648] Input: Stored training data, training goals and criteria

[1649] Output: Initialized training model

[1650] Step 4:

[1651] The server runs repeated training sessions on the learning model, having the model perform business tasks based on the training data, and evaluates its performance. After each session, the evaluation results are recorded in a database and sent to the device.

[1652] Input: Learning model, training data

[1653] Output: Evaluation results, updated learning model

[1654] Step 5:

[1655] The device analyzes the user's facial expressions and voice in real time through the emotion engine, generating emotion data, which is then sent to the server.

[1656] Input: User's facial expression, voice

[1657] Output: Emotion data

[1658] Step 6:

[1659] The server receives the acquired emotion data and reflects it in adjusting the performance of the learning model and the workflow. Specifically, it takes operational optimization measures such as reallocating tasks according to the worker's emotions.

[1660] Input: Emotion data, learning model

[1661] Output: Optimized workflow, adjusted learning model

[1662] Step 7:

[1663] Once training is complete and the objectives are met, the server executes the process of certifying the AI ​​as a professional, and the certified AI is deployed in a practical environment such as a logistics center.

[1664] Input: Results of completed training sessions, evaluation data

[1665] Output: Certified artificial intelligence, ready for production deployment

[1666] This allows the server, terminal, and user to work together to realize learning and optimization of general-purpose artificial intelligence based on emotional feedback, enabling efficient business operations at logistics centers.

[1667] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1668] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1669] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1670] [Fourth embodiment]

[1671] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1672] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1673] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1674] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1675] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1676] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1677] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1678] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1679] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1680] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1681] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1682] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1683] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1684] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1685] Overall system configuration

[1686] Server: The core of the system, it receives, stores, processes data, generates, trains, and evaluates learning models. The server manages multiple AI learning models and serves as a specialized education platform.

[1687] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc.

[1688] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[1689] Program processing

[1690] 1. Data upload and management

[1691] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[1692] Terminal: The user uploads the selected training data file to the server via the terminal.

[1693] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[1694] 2. Setting training goals and standards

[1695] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[1696] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[1697] Server: Receives training goals and criteria and stores them in a database.

[1698] 3. Generating and initializing the learning model

[1699] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the generated learning model and performs initial settings (e.g., setting initial parameter values, setting hyperparameters for training).

[1700] 4. Running and monitoring the training process

[1701] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[1702] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[1703] User: View the AI's performance during training on a dashboard and provide feedback as needed. If the training data or criteria need to be adjusted, send new instructions to the server.

[1704] 5. Evaluation and feedback

[1705] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[1706] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1707] 6. Certification and deployment of professional AI

[1708] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1709] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1710] Specific examples

[1711] For example, if a publisher develops a medical translation AI, the following embodiment applies:

[1712] 1. Data upload:

[1713] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[1714] 2. Setting training goals and standards:

[1715] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[1716] 3. Generating and initializing the learning model:

[1717] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[1718] 4. Running the training process:

[1719] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1720] 5. Evaluation and feedback:

[1721] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[1722] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[1723] 6. Certification and deployment of professional AI:

[1724] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[1725] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1726] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[1727] The processing flow will be explained below.

[1728] Step 1:

[1729] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[1730] Step 2:

[1731] Device: The user uploads the selected training data file to the server via the device. An upload UI is provided to assist with file selection and transmission.

[1732] Step 3:

[1733] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the data and automatically inspects the format and content of the data before storing it.

[1734] Step 4:

[1735] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[1736] Step 5:

[1737] Terminal: Provides a UI for inputting training goals and criteria, sends the settings to the server, and displays a confirmation message to the user when the sending operation is complete.

[1738] Step 6:

[1739] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[1740] Step 7:

[1741] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[1742] Step 8:

[1743] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[1744] Step 9:

[1745] Server: Automatically evaluates the AI's performance after each training session, records the results, scores them based on the evaluation criteria, and stores them in a database.

[1746] Step 10:

[1747] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[1748] Step 11:

[1749] User: Check the AI's performance during training on the dashboard and provide feedback as needed. Feedback is sent from the device to the server.

[1750] Step 12:

[1751] Server: After each training session, the server evaluates the AI's performance and uses the feedback to retune the learning model, resetting parameters as needed.

[1752] Step 13:

[1753] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1754] Step 14:

[1755] Server: Receives new settings from the user, adjusts the training plan again, and continues the learning session.

[1756] Step 15:

[1757] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1758] Step 16:

[1759] Device: Informs users of certification information and, if necessary, assists them in deploying the AI ​​in a production environment, ensuring that the AI ​​works effectively in the field.

[1760] Example 1

[1761] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1762] Adapting general-purpose AI to specific specialized fields requires effective management of diverse training data and the generation of highly accurate learning models. However, conventional systems have not adequately verified the consistency of specialized training data, monitored the training process in real time, or applied the data to a practical environment after the training goal has been achieved. As a result, it has been difficult to efficiently and effectively develop specialized AI and deploy it in practice.

[1763] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1764] In this invention, the server includes means for uploading training data selected by industry experts, means for verifying the integrity of the uploaded training data and storing it, means for setting training goals and standards based on the field of expertise, means for generating and initializing a learning model of the general purpose AI based on the set training goals and standards, means for performing repeated learning sessions on the learning model using the training data, means for evaluating the performance of the AI ​​after the learning session and notifying the user of the results, means for adjusting the learning model based on feedback from the user, means for certifying the AI ​​as a professional when training is completed and the goal is achieved, and means for exporting the generated professional AI model for application to a practical environment. This makes it possible to efficiently and effectively adapt the general purpose AI to a specific field of expertise and to develop and deploy professional AI that is suited to practical use.

[1765] An "industry expert" is someone who has extensive knowledge and experience in a particular field and is able to select training data and set standards in that field.

[1766] "Training data" is a collection of data used to train an artificial intelligence to acquire the knowledge necessary to perform a specific task.

[1767] "Integrity check" refers to the process of verifying that uploaded data is in the correct format, free of missing or duplicated data, and suitable for training.

[1768] "Storage" refers to the technology used to hold uploaded data using a database or storage system.

[1769] "Training goals" refer to the specific performance or functional goals that artificial intelligence should achieve through training.

[1770] "Training standards" refer to the guidelines and rules necessary to achieve training goals, including standards for translation style and accuracy.

[1771] A "learning model" refers to an algorithm or neural network that artificial intelligence generates based on training data.

[1772] "Initialization" refers to the process of initializing the generated learning model and preparing it for training.

[1773] "Iterative learning sessions" refers to the learning process of using training data to update the learning model over multiple cycles to improve performance.

[1774] "Performance evaluation" refers to the process of calculating indicators to measure how close a learning model is to achieving its goals and analyzing the results.

[1775] "Means for notifying users" refers to the technology and methods for notifying users of performance evaluation results via a dashboard or notification function.

[1776] "Adjusting based on feedback" refers to the process of changing the parameters and training data of a learning model based on user opinions and evaluations to improve the accuracy of the model.

[1777] "Professional certification" refers to the process of officially recognizing artificial intelligence that has met its training objectives as suitable for use in practice in a particular professional field.

[1778] "Export means" refers to the technology or method for converting and outputting a certified AI model into an appropriate data format for application in a production environment.

[1779] "Applying to a production environment" refers to the process of actually implementing and using a certified AI model in a specific business or task.

[1780] The specialized education platform for adapting a general-purpose AI to a specific specialized field according to the present invention is mainly composed of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1781] Overall system configuration

[1782] Server: This is the core of the system and has the functions of receiving, storing, and processing data, as well as generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Suitable software for use includes MySQL for database management and TensorFlow or PyTorch for generating and training AI models.

[1783] Terminal: Provides an interface for users to operate the system. The terminal allows users to upload data, set training goals and standards, monitor the training process, provide feedback, etc. The interface is generally provided through a web browser.

[1784] Users: Industry experts who provide training data and benchmarks, as well as feedback to the AI ​​during training.

[1785] Program processing

[1786] The server processes the program as follows:

[1787] Data Upload and Management:

[1788] Users prepare training data related to their field of expertise. For example, if training a medical translation AI, they would select medical books, papers, medical records, and terminology dictionaries.

[1789] Through the device's web browser, the user uploads the selected training data file, for example, using an HTML5-based file upload widget.

[1790] The server receives uploaded training data, stores it in a MySQL database, and uses Python scripts to check the data for consistency and correctness.

[1791] Setting training goals and standards:

[1792] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI.

[1793] The terminal provides a form for transmitting the entered training goals and criteria to the server.

[1794] The server validates the received training goals and criteria and stores them in a database.

[1795] Generate and initialize the learning model:

[1796] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[1797] Running and monitoring the training process:

[1798] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, a Python script automatically evaluates the AI's performance and records the results.

[1799] The terminal provides a dashboard for monitoring the training process in real time and displays the evaluation results using JavaScript.

[1800] Users can check the AI's performance during training on a dashboard and provide feedback as needed. If necessary, they can input training data or instructions for adjusting the criteria into the device and send them to the server.

[1801] Rating and feedback:

[1802] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[1803] Based on the training results, the user modifies the training data and criteria as needed and sends the new settings to the server.

[1804] Certification and deployment of professional AI:

[1805] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[1806] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[1807] Specific examples

[1808] As an example, we will explain a specific embodiment in which a publisher develops a medical translation AI.

[1809] 1. Data upload:

[1810] User: Selects learning data such as medical books, papers, and medical records and uploads them to the server via the device.

[1811] 2. Setting training goals and standards:

[1812] User: Inputs criteria for improving the accuracy of medical translations or specific translation styles from the terminal and sends them to the server.

[1813] 3. Generating and initializing the learning model:

[1814] Server: Generates and initializes AI learning models using TensorFlow and PyTorch based on medical-related data and standards.

[1815] 4. Running the training process:

[1816] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1817] 5. Evaluation and feedback:

[1818] Server: Notifies the training results via a dashboard and adjusts the learning model as needed.

[1819] User: View the results, provide feedback, and send instructions for any necessary adjustments.

[1820] 6. Certification and deployment of professional AI:

[1821] Server: Once the AI ​​achieves its goals, it undergoes a certification process and is deployed in a production environment.

[1822] On the device: Inform the user of the authentication information and provide an expand button.

[1823] Example prompts to input to the generative AI model

[1824] Please generate a learning model specialized for medical translation AI. Using the following data, the training goal is "improving translation accuracy" and the evaluation criteria are "accuracy of terminology and matching of context." Upload data includes medical books, papers, and medical records. Please also perform the initial setup of the training data.

[1825] This will enable general-purpose AI to be specialized in specific fields, and through efficient and effective training, AI that is suited to practical applications will be developed.

[1826] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1827] Step 1: Data upload and management

[1828] The user prepares learning data related to their field of expertise (e.g., medical books, papers, medical records, glossaries). Next, they open a web browser on their device, access the system's data upload screen, click the "Select File" button, and select the learning data file they have selected. Then, they press the "Upload" button to send the data to the server.

[1829] Input: Training data files such as medical books, papers, and medical records

[1830] Data processing and calculation: Check file format, detect missing data, delete duplicate data

[1831] Output: The training data with consistency confirmed is saved in the database.

[1832] Specific operation: The server stores the uploaded file in temporary storage, runs a Python script to check the file format, check for missing data, and remove duplicates, and then stores it in the official database.

[1833] Step 2: Setting training goals and standards

[1834] Users input the AI's training goals (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) through the device's UI. Users enter the training goals and criteria in the text fields and execute the settings by pressing the save button.

[1835] Input: Training goals and standards (e.g., "Achieve medical translation accuracy of 95% or higher," "Use technical terms preferentially")

[1836] Data processing and data calculation: Validation of input data (format confirmation and content consistency confirmation)

[1837] Output: Training goals and criteria are stored in a database

[1838] Specific behavior: The server checks the received training goals and criteria with validation scripts (e.g., checking for invalid input), executes SQL queries, and records them in the database.

[1839] Step 3: Generate and initialize the learning model

[1840] The server generates a general-purpose AI learning model using TensorFlow or PyTorch based on the learning data and training criteria received from the user. The generated learning model is initialized and the necessary parameters are set.

[1841] Input: Consistent training data, training criteria

[1842] Data processing and data calculation: Defining model architecture, initializing parameters, setting hyperparameters

[1843] Output: Initialized AI learning model

[1844] What happens: The server runs a Python script to define the layers of the neural network and set the initial parameters, so the learning model is ready to start training.

[1845] Step 4: Running and monitoring the training process

[1846] The server runs repeated training sessions on the learning model. During each session, the AI ​​performs a specific task (e.g., translating medical documents) based on the training data. After each session, the AI's performance is automatically evaluated and the results are recorded.

[1847] Input: Consistency-checked training data, initialized AI learning model

[1848] Data processing and calculation: data batching, error calculation, backpropagation, model updating

[1849] Output: Performance evaluation results after each session

[1850] Specific operation: The server divides the training data into batches and inputs them into the model, calculates the error function for each batch, performs backpropagation to update the model, and records the training progress and evaluation results in a log file.

[1851] Step 5: Evaluate and incorporate feedback

[1852] The server evaluates the AI's performance after each training session and displays the results on a dashboard. It also receives user feedback and incorporates it into the learning model.

[1853] Input: performance results after training sessions, user feedback

[1854] Data processing and data calculation: Calculation of evaluation metrics (e.g., precision, recall, F-measure), analysis of feedback, model retraining

[1855] Output: Training results displayed in a dashboard, with an updated model incorporating feedback

[1856] Specific operation: The server executes the evaluation script to evaluate the model's performance, records the results in a database, and displays them on a dashboard. It also performs retraining and parameter adjustment based on user feedback.

[1857] Step 6: Certification and deployment of professional AI

[1858] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in ONNX format, and prepares it for deployment in a production environment.

[1859] Input: Completely trained AI model, training result data

[1860] Data processing and calculation: Final adjustment of the model, export of the model

[1861] Output: AI models exported in a format suitable for production environments

[1862] Specific operation: The server runs an auxiliary script to evaluate the training results, and if it determines that the goal has been achieved, it exports the model in ONNX format and packages it into a Docker container.

[1863] Step 7: Notification of certification information and deployment of AI

[1864] The terminal will notify the user of the certification information and, if necessary, provide a deployment button, which the user can click to deploy the AI ​​model to a production environment.

[1865] Input: Certified AI model, user deployment instructions

[1866] Data processing and data calculation: Notification of certification information, settings for deployment

[1867] Output: AI model deployed in production

[1868] Specific operation: The user checks the notification on the device and clicks the deploy button to deploy the AI ​​model to the production environment. The server receives the instruction and actually performs the deployment.

[1869] (Application example 1)

[1870] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1871] Conventional methods for training and evaluating AI robots specialized for specific tasks in factories are time-consuming and costly, preventing efficient automation. Furthermore, it is difficult to monitor the training process and check evaluation results in real time, making it difficult to provide appropriate feedback in a timely manner. A method for solving these problems and developing AI robots specialized in specific fields is needed.

[1872] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1873] In this invention, the server includes: means for uploading training data selected by industry experts; means for verifying the integrity of the uploaded training data and storing it; means for setting training goals and standards based on the field of expertise; means for generating and initializing a learning model for a general AI based on the set training goals and standards; means for performing repeated learning sessions on the learning model using the training data; means for evaluating the performance of the AI ​​after the learning sessions and notifying a user of the results; means for adjusting the learning model based on user feedback; means for certifying the AI ​​as a professional when training is complete and the goals are achieved; means for training an AI robot specialized in a specific task to automate work in a factory; and means for setting training data and training goals using a smartphone and checking the training process and evaluation results in real time, which not only significantly improves work efficiency in the factory and reduces time and costs, but also enables real-time monitoring of the training process and checking the evaluation results.

[1874] "General purpose artificial intelligence" refers to artificial intelligence that has a wide range of applications, not limited to specific tasks or fields.

[1875] A "specialized education platform" is a system for training and educating artificial intelligence specialized in specific fields.

[1876] An "industry expert" is an expert with advanced knowledge and experience in a particular field of expertise.

[1877] "Training data" refers to the data set used to train an artificial intelligence.

[1878] "Upload" refers to the act of transferring data from a local terminal to a server.

[1879] "Integrity" means that data is correct, consistent, and free from errors.

[1880] "Preservation" refers to the act of storing data for a long period of time.

[1881] A "training goal" is a specific goal that you want to achieve in learning artificial intelligence.

[1882] "Training standards" refer to the standards and rules that artificial intelligence must follow during the training process.

[1883] A "learning model" is an artificial intelligence algorithm built based on training data.

[1884] "Initialization" refers to the act of resetting the parameters of a learning model to their initial settings.

[1885] A "learning session" is the process by which an artificial intelligence repeatedly learns using training data.

[1886] "Performance evaluation" refers to evaluating the performance of a trained artificial intelligence.

[1887] "User" refers to a person or institution using the system.

[1888] "Feedback" refers to evaluations and comments on training results provided by users.

[1889] "Accreditation" is the act of officially recognizing a qualification or status based on specific criteria.

[1890] A "professional" is a person who has advanced knowledge and skills in a specific specialized field and is engaged in that occupation, or a system that supports such a person.

[1891] "Factory work" refers to a series of tasks and processes carried out on the factory production floor.

[1892] "Automation" is the act of enabling machines or systems to perform tasks autonomously without human intervention.

[1893] A "smartphone" is a mobile phone that has computing capabilities.

[1894] "Real time" means that events are processed and displayed as they occur.

[1895] A "dashboard" is a user interface that visually displays the status and data of a system.

[1896] The professional education platform of the present invention is designed for the purpose of training artificial intelligence robots to automate specific tasks in factories. This system mainly consists of three elements: a server, a terminal, and a user. A specific embodiment of this system will be described below.

[1897] Overall system configuration

[1898] server

[1899] The server is the core of this system and has the functions of receiving, storing, and processing data, generating, training, and evaluating learning models. The server manages multiple AI learning models and serves as a specialized education platform. Specifically, the server uses the following hardware and software:

[1900] Hardware: High performance computer, GPU (e.g. NVIDIA GPU)

[1901] Software: Python, Flask, PyTorch, Transformers library

[1902] Terminal

[1903] The terminal provides an interface for users to operate the system. Smartphones are mainly used. The terminal provides the following functions:

[1904] Uploading data

[1905] Setting training goals and standards

[1906] Monitoring the training process

[1907] Providing Feedback

[1908] User

[1909] Users are experts in factory operations and provide training data and standards, as well as feedback to the AI ​​during training.

[1910] Program processing

[1911] Data Upload and Management

[1912] User: The user prepares training data related to factory operations, such as product inspection images and sensor data.

[1913] Terminal: Upload the selected training data file to the server via the terminal.

[1914] Server: Receives uploaded training data, stores it in a database, and checks the integrity and accuracy of the data.

[1915] Setting training goals and standards

[1916] User: The user inputs the AI ​​training goal (e.g., improving the accuracy of product inspection) and training criteria (e.g., inspection speed, false positive rate) from the terminal.

[1917] Terminal: Provides a UI for inputting training goals and criteria, and sends the settings to the server.

[1918] Server: Receives training goals and criteria and stores them in a database.

[1919] Generating and initializing the learning model

[1920] Server: Generates a general-purpose AI learning model based on the training data and standards received from the user. Initializes the generated learning model and performs initial settings.

[1921] Running and monitoring the training process

[1922] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specific task (e.g., product inspection) based on the training data. The server automatically evaluates the AI's performance after each training session and records the results.

[1923] Terminal: Allows real-time monitoring of the training process and displays evaluation results on a dashboard.

[1924] Evaluation and feedback

[1925] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. If necessary, readjust the learning model based on user feedback.

[1926] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1927] Certification and deployment of professional AI

[1928] Server: Once it determines that the training objectives have been achieved, it executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1929] Terminal: Informs users of certification information and deploys AI in production environments as needed.

[1930] Specific examples

[1931] As an example, consider an AI robot that automates the task of inspecting products on a conveyor belt in a factory. First, the user uploads the image data of the product inspection to a server using a smartphone. Next, the user inputs training goals and criteria for the AI ​​to achieve an accuracy of 95% or more in product inspection. Once the training process is complete, the user can view the evaluation results on a dashboard and provide feedback.

[1932] Example prompt sentence:

[1933] "Start the Factory Robot AI Trainer app and upload the image data of the object to be inspected. Then, enter the training goals and criteria for the AI ​​robot to achieve a product inspection accuracy of 95% or more. After training is complete, review the evaluation results and provide feedback."

[1934] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1935] Step 1:

[1936] Data upload

[1937] Users prepare training data related to factory operations (e.g., product inspection images and sensor data) and upload it to the server via their terminal. The terminal checks the file format and size of the uploaded data and transfers it to the server. The server receives the file and stores it in a database. The input is the training data, and the output is the data stored in the server's database.

[1938] Step 2:

[1939] Setting training goals and standards

[1940] The user inputs training goals (e.g., improving product inspection accuracy) and training criteria (e.g., inspection speed, false positive rate) from the terminal. The terminal sends the goals and criteria entered by the user to the server. The server receives this data and stores it in a database. The input is the goals and criteria entered by the user, and the output is the training goals and criteria stored in the server's database.

[1941] Step 3:

[1942] Generating and initializing the learning model

[1943] The server generates and initializes a general-purpose AI learning model based on the uploaded training data and the set goals and criteria. Specifically, it uses Python, PyTorch, and the Transformers library to build the learning model and set initial parameters. The input is the training data, goals, and criteria, and the output is the initialized learning model.

[1944] Step 4:

[1945] Running and monitoring the learning process

[1946] The server runs repeated learning sessions on the generated learning model. Specifically, the AI ​​performs a specific task (e.g., product inspection) using the uploaded data and records the results. After each session, the server evaluates the AI's performance and records the results. The device monitors this process in real time and displays the evaluation results on a user interface (dashboard). The inputs are the training data and the learning model, and the output is the evaluation results.

[1947] Step 5:

[1948] Evaluation and feedback

[1949] The server evaluates the AI's performance after each training session and displays the evaluation results on a dashboard. The user can check the evaluation results on the dashboard and provide feedback as needed. The server receives feedback from the user and readjusts the training model. The inputs are the evaluation results and user feedback, and the output is the adjusted training model.

[1950] Step 6:

[1951] Certification and deployment of professional AI

[1952] If the server determines that the training goals have been achieved, it executes the process of certifying the AI ​​as a professional AI. The certified AI model is exported in an appropriate format and prepared for deployment in a production environment. The terminal notifies the user of the certification information and deploys the AI ​​in a production environment as necessary. The input is the achievement status of the training goals, and the output is the certified AI model and notification.

[1953] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1954] The specialized training platform for adapting general-purpose AI to specific specialized fields according to the present invention is composed of three elements: a server, a terminal, and a user. Furthermore, the system combines an emotion engine that recognizes user emotions and utilizes user emotional feedback in the AI ​​training process.

[1955] Overall system configuration

[1956] Server: The core of the system, it receives, stores, and processes data, generates and trains learning models, evaluates them, and processes the emotion engine.

[1957] Terminal: Provides an interface for users to operate the system and has emotion recognition functionality using an emotion engine.

[1958] Users: Industry experts who provide training data and standards, and emotional feedback from the emotion engine.

[1959] Emotion engine: Analyzes the user's facial expressions, voice, and input to detect emotions and reflect them in performance and feedback.

[1960] Program processing

[1961] 1. Data upload and management

[1962] User: Prepare training data related to their field of expertise. For example, if training a medical translation AI, select medical books, papers, medical records, and terminology dictionaries.

[1963] Terminal: The user uploads the selected training data file to the server through the terminal. The terminal interface assists the upload operation.

[1964] Server: Receives uploaded training data and stores it in the database. Checks the integrity and correctness of the data and stores it in the database.

[1965] 2. Setting training goals and standards

[1966] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context) from the device.

[1967] Terminal: Provides an interface for inputting training goals and criteria and transmits the settings to the server.

[1968] Server: Receives training goals and criteria and stores them in a database. Creates training plans based on the received settings.

[1969] 3. Generating and initializing the learning model

[1970] Server: Generates a general-purpose AI learning model based on the learning data and training criteria received from the user. Initializes the model and sets parameters to prepare for training.

[1971] 4. Running and monitoring the training process

[1972] Server: Runs repeated training sessions on the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data. After each training session, the AI's performance is automatically evaluated and the results are recorded.

[1973] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to check the current training progress.

[1974] 5. Processing emotional feedback

[1975] Device: The emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The emotion recognition results are sent to the server.

[1976] Server: Receives user emotion data and uses it to adjust learning models and optimize feedback responses.

[1977] 6. Evaluation and feedback

[1978] Server: After each training session, evaluate the AI's performance. Display the results on a dashboard and notify the user. Readjust the learning model based on the user's emotional data.

[1979] User: Based on the training results, modify the training data and criteria as needed and send the new settings to the server.

[1980] 7. Certification and deployment of professional AI

[1981] Server: Once the training objectives are met, the server executes the process of certifying the AI ​​as a professional AI, exports the certified AI model in an appropriate format, and prepares it for deployment in a production environment.

[1982] Terminal: Informs users of certification information and assists them in deploying AI in production environments as needed.

[1983] Specific examples

[1984] For example, if a publisher develops a medical translation AI, the following embodiment applies.

[1985] 1. Data upload:

[1986] User: Selects medical books, papers, medical records, etc. and uploads them to the server via the terminal.

[1987] 2. Setting training goals and standards:

[1988] User: Enter the goals for improving the accuracy of AI medical translation and translation style standards on the device, and send the settings to the server.

[1989] 3. Generating and initializing the learning model:

[1990] Server: Generates and initializes the AI ​​learning model based on medical-related data and standards.

[1991] 4. Running the training process:

[1992] Server: Trains the translation of medical documents based on the training data and evaluates performance after each session.

[1993] 5. Processing emotional feedback:

[1994] Terminal: Recognizes the user's emotions in real time during training and sends the results to the server via the emotion engine.

[1995] Server: Leverages emotional data to adjust learning models and feedback responses.

[1996] 6. Evaluation and Feedback:

[1997] Server: Evaluates the training results and retunes the model based on user emotional feedback.

[1998] User: Modify the training data or criteria as needed and resubmit.

[1999] 7. Certification and deployment of professional AI:

[2000] Server: Executes the certification process when training objectives are achieved and deploys to the production environment.

[2001] Terminal: Notifies certification information and assists with AI deployment.

[2002] This allows the system to efficiently and effectively specialize general AI in specific fields of expertise, and comprehensively utilize user feedback to develop high-quality professional AI.

[2003] The processing flow will be explained below.

[2004] Step 1:

[2005] Users: Prepare training data related to their field of expertise. For example, if training a medical translation AI, users would select medical books, papers, medical records, and terminology dictionaries.

[2006] Step 2:

[2007] Terminal: The user uses an interface to upload the selected training data file to the server through the terminal. The user is assisted in selecting the file and pressing the upload button.

[2008] Step 3:

[2009] Server: Receives uploaded training data and stores it in a database. It checks the integrity and correctness of the uploaded data and inspects the format and content for any problems.

[2010] Step 4:

[2011] User: Inputs the AI ​​training goal (e.g., improving the accuracy of medical translation) and training criteria (e.g., translation style, context selection) from the terminal.

[2012] Step 5:

[2013] Terminal: Provides an interface for inputting training goals and criteria, transmits the settings to the server, and displays a confirmation message to the user upon completion.

[2014] Step 6:

[2015] Server: Receives training goals and criteria and stores them in a database. Creates a training plan based on the received settings information.

[2016] Step 7:

[2017] Server: Generates and initializes a general-purpose AI learning model based on the learning data and training criteria received from the user. Initialization involves setting model parameters and loading initial data.

[2018] Step 8:

[2019] Server: Runs repeated training sessions against the learning model. In each session, the AI ​​performs a specified task (e.g., translating medical documents) based on the training data.

[2020] Step 9:

[2021] Server: After each training session, the server automatically evaluates the AI's performance and records the results. The evaluation is based on set criteria and saved in the form of a score.

[2022] Step 10:

[2023] Terminal: Enables real-time monitoring of the training process and displays evaluation results on a dashboard, allowing users to see training progress and performance.

[2024] Step 11:

[2025] Device: During training, the emotion engine analyzes the user's facial expressions, voice, and input content in real time to detect emotions. The detected emotion data is sent to the ser...

Claims

1. A specialized education platform for adapting general artificial intelligence to specific specialized fields, a means for uploading training data selected by industry experts; a means of verifying and storing the integrity of the uploaded training data; a means of establishing discipline-based training goals and standards; A means for generating and initializing a learning model of the general artificial intelligence based on set training goals and criteria; means for performing repeated training sessions on the learning model using training data; a means for evaluating the performance of the artificial intelligence after a training session and informing the user of the results; a means for adjusting the learning model based on user feedback; and A means to certify AI as a profession once training is complete and goals are met; and A system including:

2. The system of claim 1 , further comprising means for monitoring the training process of the artificial intelligence in real time and displaying the evaluation results on a dashboard.

3. 2. The system of claim 1, wherein the training data based on a specific field of expertise are medical-related documents, and the learning model is specialized for medical translation.

Images