Artificial intelligence application development system, computer equipment and storage medium

Abstract

The invention belongs to the technical field of artificial intelligence, and provides an artificial intelligence application development system, computer equipment and a storage medium, and the system comprises a data processing module which collects data from various sources, and carries out the preprocessing cloud storage of the data, and obtains a training data set; the model construction module is used for constructing and initializing an AI model according to image recognition and natural language processing in combination with data characteristics, and training the constructed AI model by using a training data set; the model deployment module converts the trained AI model into AP I or micro-service; the prediction reasoning module receives user input, returns a model prediction result, processes batch prediction requests, and provides a prediction result and confidence; the monitoring feedback module monitors the model performance in real time; the user interaction module provides an interaction interface, so that a user obtains required information and service; according to the method, the development and maintenance efficiency can be improved, the software quality is improved, the user experience is optimized, and innovation and commercial value increase are promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence, and specifically relates to an artificial intelligence application development system, a computer device, and a storage medium. Background Art

[0002] In recent years, artificial intelligence (AI) technology has developed rapidly and gradually penetrated into multiple industries, promoting the wide application of intelligent applications;

[0003] However, most current artificial intelligence application development tools are often highly specialized for specific purposes or algorithms. This requires developers to have in-depth domain knowledge, especially in data science, machine learning, and deep learning, which increases the threshold for their use. Most existing systems lack an intuitive user interface and friendly interaction design. Non-professionals often face setbacks when trying to use existing tools and cannot effectively start and run their own AI projects. Existing development frameworks mostly rely on manual code writing and debugging, and the workflow is often cumbersome and time-consuming. Although there are some automated tools attempting to alleviate this burden, they are usually only applicable to specific development links and do not cover the entire development life cycle, from data preprocessing to model deployment, lacking systematic support, which makes it easy to encounter bottlenecks during the development process. In addition, this manual operation is prone to errors, increasing the complexity of model development and iteration. Moreover, with the continuous increase in the amount of data, existing AI development tools often perform poorly when dealing with large-scale datasets. Most systems rely on traditional data processing architectures and are difficult to fully utilize the advantages of modern distributed computing and cloud computing, resulting in waste of data resources and affecting the efficiency of model training and inference.

[0004] Therefore, those skilled in the art have proposed an artificial intelligence application development system, a computer device, and a storage medium, aiming to improve the efficiency of development and maintenance, enhance software quality, optimize the user experience, and promote innovation and business value growth. Summary of the Invention

[0005] To solve the above technical problems, the present invention provides an artificial intelligence application development system, a computer device, and a storage medium to solve the problems raised in the background art.

[0006] According to a first aspect of the present disclosure, an artificial intelligence application development system is proposed, including:

[0007] A data processing module for collecting data from various sources, including sensors, databases, file systems, web crawlers, providing raw data, and preprocessing the data, including data cleaning and data transformation, and storing the preprocessed data in the cloud to obtain a training dataset;

[0008] The model construction module is used to construct and initialize an AI model according to the combination of image recognition and natural language processing with data characteristics, and train the constructed AI model using a training data set;

[0009] The model deployment module is used to convert the trained AI model into an API or microservice;

[0010] The prediction and inference module is used to receive user input and return the model prediction result, process batch prediction requests, and provide the prediction result and confidence level;

[0011] The monitoring and feedback module is used to monitor the model performance in real time and record user feedback data;

[0012] The user interaction module is used to provide an interaction interface for the user to obtain the required information and services.

[0013] Preferably, the data processing module obtains data from sensors, databases, file systems, and web crawler data sources, and preprocesses the data, including data cleaning and data conversion. The data cleaning includes removing noise, filling missing values, detecting and processing outliers. The data conversion is to convert the data into a format for AI model training; and store the preprocessed data in a cloud database to obtain a training data set.

[0014] Preferably, the model construction module constructs a model based on image recognition, obtains an image data set from the data processing module, and constructs it using a convolutional neural network. The input layer input size is (H, W, C), where H is the image height, W is the width, and C is the number of channels. Spatial features are extracted through the convolutional layer, and the dimension is reduced by the pooling layer. The fully connected layer makes classification decisions. The convolutional operation is expressed as:

[0015]

[0016] where I represents the input image, K represents the convolutional kernel, (I * K)[i, j] represents the value of the result of the convolutional operation at position (i, j), m and n respectively traverse the row and column indices of the input image area covered by the convolutional kernel, and K[i - m, j - n] represents the value of the convolutional kernel K at the offset position (i - m, j - n) relative to the current output position (i, j);

[0017] The model construction module constructs a model based on natural language processing and constructs it using a recurrent neural network. Its Embedding layer converts words into dense vectors, the RNN layer processes sequence data, and the fully connected layer serves as the output layer for classification or generation tasks. The RNN output calculation is expressed as:

[0018] h t = f(W hh t-1 +W x x t +b)

[0019] Among them, x t represents the current input, h t represents the current hidden state, W h ,W x respectively represent the weight matrices, and b represents the bias.

[0020] Preferably, the AI model constructed by the model construction module based on image recognition and natural language processing uses the cross-entropy loss as its loss function, which is expressed as:

[0021]

[0022] Among them, y i represents the true label of the i-th sample, represents the probability value predicted by the model for the i-th sample, N represents the total number of samples, and i represents the index;

[0023] According to the characteristics of the data combined with image recognition and natural language processing, a structured data model is constructed, and the gradient boosting of the mean squared error is used. The objective function is expressed as:

[0024]

[0025] Among them, Ω(f k ) is the regularization term used to ensure the complexity of the model. K represents the number of base learners in the model, and f k represents the k-th base learner;

[0026] For the structured data model constructed according to image recognition and natural language processing and combined with data characteristics, the stochastic gradient descent optimization algorithm is adopted to update the weights to minimize the loss function, which is expressed as:

[0027]

[0028] Among them, w is the weight of the model, η is the learning rate, L is the loss function, and the training data set obtained by the data processing module is used for model training.

[0029] Preferably, the prediction and inference module sends the user's input through the API interface, receives and processes a single prediction request or a batch prediction request for the input. For the input data of a single sample, it is expressed as: x = [x1, x2,..., x n , and for the input data of a batch of samples, it is expressed as:

[0030]

[0031] Among them, m is the batch size and n is the number of features;

[0032] When using this AI model for inference and mapping the input to the predicted output, then:

[0033]

[0034] Among them, is the predicted output, θ is the model parameter, and f is the functional representation of the model; for multi-class classification problems, the output probability of the model is:

[0035]

[0036] Among them, z k is the unnormalized score of the AI model for class k, K is the number of classes, and p(y = k|X) represents the probability that the model predicts the output as class k given the input feature vector X. represents the exponential value of the score of class k, and j represents the index;

[0037] The confidence level is represented by the maximum value in the predicted output vector, which is expressed as:

[0038] C = max(p(y = k|x))

[0039] The predicted class is obtained and represented as: predicted-class = argmax(p(y = k|x)), where p(y = k|x) represents the probability that the model predicts the output as class k given the input feature vector x.

[0040] According to the second aspect of the present disclosure, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the artificial intelligence application development system described in the first aspect are implemented.

[0041] According to the third aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the artificial intelligence application development system described in the first aspect are implemented.

[0042] According to the fourth aspect of the present disclosure, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps of the artificial intelligence application development system described in the first aspect are implemented.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. The present invention realizes personalized recommendations by analyzing user data and behaviors, improves user satisfaction, and enhances the user interaction experience through real-time prediction and automated support services. At the same time, artificial intelligence can process large amounts of data and make decisions quickly, helping developers discover hidden information and patterns, promoting the formation of innovation, and providing valuable suggestions based on the results of data analysis.

[0045] 2. The present invention helps enterprises optimize the decision-making process and business strategies through data analysis and prediction models, and has the ability to process batch data sets. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a block diagram of an artificial intelligence application development system of the present invention.

[0047] Figure 2 It is a schematic structural diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The following further describes in detail the embodiments of the present invention in conjunction with the drawings and embodiments. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0049] As shown in the Figure 1 drawings:

[0050] Embodiment 1: The present invention provides an artificial intelligence application development system, including:

[0051] A data processing module, configured to collect data from various sources, including sensors, databases, file systems, web crawlers, provide raw data, and preprocess the data, including data cleaning and data conversion, and store the preprocessed data in the cloud to obtain a training data set;

[0052] A model construction module, configured to construct and initialize an AI model according to the combination of image recognition and natural language processing with data characteristics, and train the constructed AI model using the training data set;

[0053] A model deployment module, configured to convert the trained AI model into an API or a microservice;

[0054] A prediction and inference module, configured to receive user input and return model prediction results, process batch prediction requests, and provide prediction results and confidence levels;

[0055] A monitoring and feedback module, configured to monitor the model performance in real time and record user feedback data;

[0056] A user interaction module, configured to provide an interaction interface for users to obtain the required information and services.

[0057] The data processing module obtains data from sensors, databases, file systems, and web crawler data sources, and preprocesses the data, including data cleaning and data transformation. Data cleaning includes removing noise, filling in missing values, detecting and handling outliers. Data transformation is to convert the data into a format suitable for AI model training; and stores the preprocessed data in a cloud database to obtain a training dataset.

[0058] The model construction module constructs a model based on image recognition. It obtains an image data set from the data processing module and constructs it using a convolutional neural network. The input layer has an input size of (H, W, C), where H is the image height, W is the width, and C is the number of channels. Spatial features are extracted through convolutional layers, and the dimension is reduced by pooling layers. The fully connected layer makes classification decisions. The convolutional operation is expressed as:

[0059]

[0060] where I represents the input image, K represents the convolutional kernel, (I * K)[i, j] represents the value of the result of the convolutional operation at position (i, j), m and n respectively traverse the row and column indices of the area of the input image covered by the convolutional kernel, and K[i - m, j - n] represents the value of the convolutional kernel K at the offset position (i - m, j - n) relative to the current output position (i, j);

[0061] The model construction module constructs a model based on natural language processing and uses a recurrent neural network for construction. Its Embedding layer converts words into dense vectors, the RNN layer processes sequence data, and the fully connected layer serves as the output layer for classification or generation tasks. The RNN output calculation is expressed as:

[0062] h t =f(W h h t-1 +W x x t +b)

[0063] where x t represents the current input, h t represents the current hidden state, W h , W x represent weight matrices respectively, and b represents the bias. In the model construction module, constructing and training an AI model according to the characteristics of different tasks such as image recognition, natural language processing, and structured data processing is the key to realizing an effective artificial intelligence system. By constructing the corresponding AI model and using the training dataset for effective training.

[0064] For the AI models constructed by the model construction module based on image recognition and natural language processing, their loss functions both use cross-entropy loss, which is expressed as:

[0065]

[0066] Among them, y i represents the true label of the i-th sample, represents the probability value predicted by the model for the i-th sample, N represents the total number of samples, and i represents the index; the loss function is used to evaluate the gap between the model's prediction and the true label.

[0067] According to the characteristics of the combined data of image recognition and natural language processing, a structured data model is constructed, and gradient boosting with squared error is used. The objective function is expressed as:

[0068]

[0069] Among them, Ω(f k ) is the regularization term, which is used to ensure the complexity of the model. K represents the number of base learners in the model, and f k represents the k-th base learner;

[0070] For the structured data model constructed according to image recognition, natural language processing, and combined with data characteristics, the stochastic gradient descent optimization algorithm is adopted to update the weights to minimize the loss function, which is expressed as:

[0071]

[0072] Among them, w is the weight of the model, η is the learning rate, L is the loss function, and the training data set obtained by the data processing module is used for model training. The model construction module preprocesses the data and performs feature engineering, and finally uses the training data set for parameter optimization to ensure that the model can effectively predict new data.

[0073] The prediction and inference module sends the user's input through the API interface, receives and processes a single prediction request or a batch prediction request for the input. For the input data of a single sample, it is expressed as: x = [x1, x2,..., x n , and for the input data of a batch of samples, it is expressed as:

[0074]

[0075] Among them, m is the batch size and n is the number of features;

[0076] Using this AI model for inference, mapping the input to the prediction output, then:

[0077]

[0078] Among them, For the predicted output, θ is the model parameter, and f is the functional representation of the model; for multi-class classification problems, the output probability of the model is:

[0079]

[0080] where z k is the unnormalized score of the AI model for class k, K is the number of classes, and p(y = k|X) represents the probability that the model predicts the output as class k given the input feature vector X. represents the exponential value of the score for class k, and j represents the index;

[0081] The confidence is represented by the maximum value in the predicted output vector, which is expressed as:

[0082] C = max(p(y = k|x))

[0083] The predicted class is obtained as: predicted_class = argmax(p(y = k|x)), where p(y = k|x) represents the probability that the model predicts the output as class k given the input feature vector x.

[0084] In the prediction inference module, the process of handling user input, returning the model prediction results, and handling batch requests is relatively straightforward. Through model inference, confidence calculation, and prediction of the class, the API can efficiently respond to and handle various input scenarios, enabling the AI model to not only perform real-time inference but also adapt to different numbers of requests, thereby improving its usability in practical applications.

[0085] As shown in the appendix Figure 2 Example 2: The present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the artificial intelligence application development system in Example 1 are implemented.

[0086] Example 3: The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the artificial intelligence application development system in Example 1 are implemented.

[0087] Example 3: The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the artificial intelligence application development system in Example 1 are implemented.

[0088] Importantly, it should be noted that the construction and arrangement of the present application shown in multiple different exemplary embodiments are merely illustrative. Although only a few embodiments are described in detail in this disclosure, those who refer to this disclosure should easily understand that many modifications are possible without materially departing from the novel teachings and advantages of the subject matter described in this application. Other substitutions, modifications, changes, and omissions may be made in the design, operating conditions, and arrangement of the exemplary embodiments without departing from the scope of the present invention. Therefore, the present invention is not limited to a specific embodiment, but extends to various modifications that still fall within the scope of the appended claims.

[0089] In addition, in order to provide a concise description of the exemplary embodiments, all features of the actual embodiments may not be described (i.e., those features that are not relevant to the best mode of implementing the present invention currently considered or those features that are not relevant to implementing the present invention).

[0090] It should be understood that in the development of any actual implementation, as in any engineering or design project, a large number of specific implementation decisions may be made. Such development efforts may be complex and time-consuming, but for those of ordinary skill in the art who benefit from this disclosure, without excessive experimentation, such development efforts will be a routine task of design, manufacturing, and production.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An artificial intelligence application development system, characterized in that: include: The data processing module is used to collect data from various sources, including sensors, databases, file systems, and web crawlers, provide raw data, and pre-process the data, including data cleaning and data conversion, and store the pre-processed data in the cloud to obtain a training data set; The model building module is used to build and initialize the AI ​​model based on image recognition and natural language processing combined with data characteristics, and train the constructed AI model using the training data set; Model deployment module, used to convert trained AI models into APIs or microservices; The prediction and reasoning module is used to receive user input and return model prediction results, process batch prediction requests, and provide prediction results and confidence levels; Monitoring feedback module, used to monitor model performance in real time and record user feedback data; The user interaction module is used to provide an interactive interface so that users can obtain the information and services they need.

2. An artificial intelligence application development system as claimed in claim 1, characterized in that: The data processing module obtains data from sensors, databases, file systems and web crawler data sources, and preprocesses the data, including data cleaning and data conversion. The data cleaning includes noise removal, missing value filling, outlier detection and processing. The data conversion is to convert the data into a format for AI model training; and the preprocessed data is stored in a cloud database to obtain a training data set.

3. An artificial intelligence application development system as claimed in claim 1, characterized in that: The model building module builds a model based on image recognition, obtains an image data set from the data processing module, and uses a convolutional neural network to build it. The input layer input size is (H, W, C), where H is the image height, W is the width, and C is the number of channels. The spatial features are proposed through the convolution layer, and the dimensions are reduced by the pooling layer. The fully connected layer performs classification decisions. The convolution operation is expressed as: Where I represents the input image, K represents the convolution kernel, (I*K)[i,j] represents the value of the result of the convolution operation at position (i,j), m and n respectively traverse the row and column indexes of the input image area covered by the convolution kernel, and K[im,jn] represents the value of the convolution kernel K at the offset position (im,jn) relative to the current output position (i,j); The model building module builds a model based on natural language processing and adopts a recurrent neural network. Its Embedding layer converts vocabulary into dense vectors, the RNN layer processes sequence data, and the fully connected layer is used as the output layer to perform classification or generation tasks. The RNN output calculation is expressed as: h t =f(W h h t-1 +W x x t +b) Among them, x t Indicates the current input, h t represents the current hidden state, W h ,W x They represent weight matrices, and b represents bias.

4. An artificial intelligence application development system as claimed in claim 3, characterized in that: The model building module constructs an AI model based on image recognition and natural language processing, and its loss function adopts cross entropy loss, which is expressed as: Among them, y i represents the true label of the i-th sample, represents the probability value predicted by the model for the i-th sample, N represents the total number of samples, and i represents the index; According to the characteristics of image recognition and natural language processing combined with data, a structured data model is constructed, and the gradient of square error is used for boosting. The objective function is expressed as: Among them, Ω(f k ) is a regular term used to ensure the complexity of the model, K represents the number of base learners in the model, and f k represents the kth base learner; The stochastic gradient descent optimization algorithm is used to update the weights of the structured data model based on image recognition and natural language processing, as well as the data characteristics, to minimize the loss function, which is expressed as: Among them, w is the weight of the model, η is the learning rate, L is the loss function, and the training data set obtained by the data processing module is used for model training.

5. An artificial intelligence application development system as claimed in claim 1, characterized in that: The prediction and inference module sends the user's input through the API interface, receives and processes the input single prediction request or batch prediction request. The input data of a single sample is expressed as: x = [x1, x2, ..., x n ], the input data of the batch sample is expressed as: Where m is the batch size and n is the number of features; Using this AI model for reasoning, the input is mapped to the predicted output, then: in, is the predicted output, θ is the model parameter, and f is the function representation of the model; for multi-category classification problems, the output probability of the model is: Among them, z k is the unnormalized score of the AI ​​model for category k, K is the number of categories, p(y=k|X) represents the probability that the model predicts the output to be category k given the input feature vector X, represents the index value of the score of category k, j represents the index; The confidence is expressed by predicting the maximum value in the output vector, which is expressed as: C = max(p(y = k | x)) The predicted category is expressed as: predicted_class = argmax(p(y = k|x)), where p(y = k|x) represents the probability that the model predicts the output to be category k given the input feature vector x.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the artificial intelligence application development system according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the artificial intelligence application development system according to any one of claims 1 to 5 are implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the artificial intelligence application development system described in any one of claims 1 to 5 are implemented.

Images