Server for providing service for educating english and method for operation thereof

Abstract

According to various embodiments, a server providing an artificial intelligence model trained according to a meta-learning method may include: a data collection unit (501) that collects data from multiple servers using web crawling and converts the collected data into numerical data; a data preprocessing unit (502) that generates training data by preprocessing the numerical data to normalize the data and performing a dimension transformation on the normalized data; a data visualization unit (503) that visualizes the training data; a model training unit (504) that trains an artificial intelligence model using the training data according to a meta-learning method; a model verification unit (505) that verifies the reliability of the artificial intelligence model by inputting verification / evaluation data corresponding to the training data into the artificial intelligence model; a model providing unit (506) that allows multiple users to access the artificial intelligence model when the reliability of the artificial intelligence model is above a threshold value; and a Human-Machine Interface (HMI) unit (507) that provides a graphical user interface that allows the multiple users to connect to the server and perform multiple operations related to the artificial intelligence model. Various other embodiments are also possible.

Description

Technology Field

[0001] Various embodiments of the present invention relate to a server that provides artificial intelligence models learned based on meta-learning. Background Technology

[0002] An artificial intelligence system is a computer system that implements human-level intelligence, enabling machines to learn and make judgments autonomously, and improving recognition accuracy with continued use. Artificial intelligence technology consists of machine learning (deep learning) technology, which utilizes algorithms to classify and learn the characteristics of input data autonomously, and component technologies that utilize machine learning algorithms to mimic functions such as cognition and judgment of the human brain. These component technologies may include, for example, at least one of the following: linguistic understanding technology that recognizes human language / characters; visual understanding technology that perceives objects like human vision; reasoning / prediction technology that judges information to logically infer and predict; knowledge representation technology that processes human experience information into knowledge data; and motion control technology that controls autonomous driving of vehicles or the movement of robots. In particular, visual understanding is a technology that perceives and processes objects like human vision, and includes object recognition, object tracking, image search, person recognition, scene understanding, spatial understanding, and image enhancement. The problem to be solved

[0003] With the spread of digital transformation and artificial intelligence technology, the demand for services applying AI is increasing across industries, leading to a rise in reliance on IT professionals (software developers); however, there is a shortage of developers.

[0004] Most AI platforms, when created by professional developers training models using large amounts of data, perform well on tasks related to the learned work (such as prediction), but cannot perform other similar tasks.

[0005] A server and a method of operation according to various embodiments of the present invention can provide a server that enables users participating in the platform to create artificial intelligence models without coding (LCNC, Low-Code, No-Code) and provide such services by utilizing Meta-Learning, which is an artificial intelligence learning method for creating general-purpose (general) models. means of solving the problem

[0006] According to various embodiments, a server providing an artificial intelligence model trained according to a meta-learning method may include: a data collection unit (501) that collects data from multiple servers using web crawling and converts the collected data into numerical data; a data preprocessing unit (502) that generates training data by preprocessing the numerical data to normalize the data and performing a dimension transformation on the normalized data; a data visualization unit (503) that visualizes the training data; a model training unit (504) that trains an artificial intelligence model using the training data according to a meta-learning method; a model verification unit (505) that verifies the reliability of the artificial intelligence model by inputting verification / evaluation data corresponding to the training data into the artificial intelligence model; a model providing unit (506) that allows multiple users to access the artificial intelligence model when the reliability of the artificial intelligence model is above a threshold value; and a Human-Machine Interface (HMI) unit (507) that provides a graphical user interface that allows the multiple users to connect to the server and perform multiple operations related to the artificial intelligence model. Effects of the invention

[0007] According to various embodiments of the present invention, users participating in the platform can create and use artificial intelligence models by minimizing coding work (LCNC), thereby increasing the efficiency of creating artificial intelligence models and providing an environment where even non-expert programmers can easily use artificial intelligence models. Brief explanation of the drawing

[0008] FIG. 1 illustrates a block diagram of an electronic device and network according to various embodiments of the present invention. FIG. 2 is a block diagram of an electronic device according to various embodiments. FIG. 3 is a block diagram of a program module according to various embodiments. FIG. 4 illustrates a learning process for learning a meta-learner according to various embodiments. FIG. 5 shows configurations for providing an artificial intelligence model included in a server according to various embodiments. FIG. 6 shows an embodiment in which a data collection unit operates according to various embodiments. FIG. 7 shows an embodiment in which a data preprocessing unit operates according to various embodiments. FIG. 8 shows an embodiment in which a data visualization unit operates according to various embodiments. FIG. 9 shows an embodiment in which a model verification unit operates according to various embodiments. Specific details for implementing the invention

[0009] The present invention can be widely applied to tasks where inference by a neural network is insufficient, and this can be overcome through a learning algorithm based on meta-learning. Specifically, the task includes the following situations: 1) a situation where inference must be obtained quickly and successfully with only a very small amount of data for an unlearned task; 2) a situation where successful inference must be obtained when the size of the neural network and the resolution of the neural network's connection weights are limited due to hardware or software issues.

[0010] In relation to 1) and 2) above, the present invention relates to meta-learning. Meta-learning can be defined as a learning paradigm for preparing in advance for unlearned tasks to be processed in the future when the task to be learned is not fixed. Therefore, meta-learning is a 'learn-to-learn' process in the sense of 'learning for the sake of learning.' While conventional learning involves learning strategies specialized for a fixed task, meta-learning according to the present invention must learn strategies capable of obtaining successful inference for new, unlearned tasks. In other words, it can be described as a paradigm for learning common inference strategies shared by changing tasks.

[0011] As described above, the meta-learner of the present invention uses as a performance evaluation criterion the speed at which it can respond to a new, unlearned task and obtain successful inference using only a very small number of samples of that task. Therefore, the core strategy of the meta-learner lies in designing an algorithm that learns common strategies shared by changing tasks, rapidly adapts to a given current task, and obtains successful inference.

[0012] Hereinafter, various embodiments of this document are described with reference to the accompanying drawings. The embodiments and the terms used therein are not intended to limit the technology described in this document to specific embodiments and should be understood to include various modifications, equivalents, and / or substitutions of said embodiments. In relation to the description of the drawings, similar reference numerals may be used for similar components. A singular expression may include a plural expression unless the context clearly indicates otherwise. In this document, expressions such as "A or B" or "at least one of A and / or B" may include all possible combinations of items listed together. Expressions such as "first," "second," "first," or "second" may modify said components regardless of order or importance and are used only to distinguish one component from another and do not limit said components. When it is mentioned that a certain (e.g., 1st) component is "(functionally or telecommunicationally) connected" or "connected" to another (e.g., 2nd) component, said certain component may be directly connected to said other component or connected through another component (e.g., 3rd component).

[0013] In this document, "configured to" may be used interchangeably with, depending on the context, for example, hardware- or software-wise, "suitable for," "capable of," "modified to," "made to," "capable of," or "designed to." In some cases, the expression "device configured to" may mean that the device is "capable of" in conjunction with other devices or components. For example, the phrase "processor configured to perform A, B, and C" may mean a dedicated processor for performing the corresponding operations (e.g., an embedded processor), or a general-purpose processor capable of performing the corresponding operations by executing one or more software programs stored in a memory device (e.g., a CPU or application processor).

[0014] An electronic device according to various embodiments of this document may include, for example, at least one of a smartphone, tablet PC, mobile phone, video phone, e-book reader, desktop PC, laptop PC, netbook computer, workstation, server, PDA, PMP (portable multimedia player), MP3 player, medical device, camera, or wearable device. A wearable device may include at least one of an accessory type (e.g., watch, ring, bracelet, anklet, necklace, glasses, contact lens, or head-mounted device (HMD)), a fabric or clothing integrated type (e.g., electronic clothing), a body-attached type (e.g., skin pad or tattoo), or a bio-implantable circuit. In some embodiments, the electronic device may include, for example, a television, DVD (digital video disk) player, audio, refrigerator, air conditioner, vacuum cleaner, oven, microwave, washing machine, air purifier, set-top box, home automation control panel, security control panel, media box (e.g., Samsung HomeSync). TM , Apple TV TM , or Google TV TM ), game console (e.g., Xbox) TM PlayStation TM It may include at least one of an electronic dictionary, an electronic key, a camcorder, or an electronic photo frame.

[0015] Referring to FIG. 1, an electronic device (101) within a network environment (100) in various embodiments is described. The electronic device (101) may include a bus (110), a processor (120), a memory (130), an input / output interface (150), a display (160), and a communication interface (170). In some embodiments, the electronic device (101) may omit at least one of the components or additionally include other components. The bus (110) may include a circuit that connects the components (110-170) to each other and transmits communication (e.g., control messages or data) between the components. The processor (120) may include one or more of a central processing unit, an application processor, or a communication processor (CP). The processor (120) may, for example, perform operations or data processing regarding the control and / or communication of at least one other component of the electronic device (101).

[0016] Memory (130) may include volatile and / or non-volatile memory. Memory (130) may store instructions or data related to at least one other component of the electronic device (101), for example. According to one embodiment, memory (130) may store software and / or programs (140). Programs (140) may include, for example, a kernel (141), middleware (143), an application programming interface (API) (145), and / or application programs (or "applications") (147), etc. At least part of the kernel (141), middleware (143), or API (145) may be referred to as an operating system. The kernel (141) can control or manage system resources (e.g., bus (110), processor (120), or memory (130), etc.) used to execute operations or functions implemented in other programs (e.g., middleware (143), API (145), or application program (147)). Additionally, the kernel (141) can provide an interface that controls or manages system resources by accessing individual components of the electronic device (101) from the middleware (143), API (145), or application program (147).

[0017] Middleware (143) may act as an intermediary to enable, for example, an API (145) or an application program (147) to communicate with the kernel (141) to exchange data. Additionally, middleware (143) may process one or more work requests received from the application program (147) according to priority. For example, middleware (143) may process the one or more work requests by assigning a priority to at least one of the application programs (147) to use the system resources of the electronic device (101) (e.g., bus (110), processor (120), or memory (130), etc.). The API (145) is an interface for the application (147) to control functions provided by the kernel (141) or middleware (143), and may include at least one interface or function (e.g., command) for, for example, file control, window control, image processing, or character control. The input / output interface (150) can, for example, transmit commands or data input from a user or other external device to other component(s) of the electronic device (101), or output commands or data received from other component(s) of the electronic device (101) to the user or other external device.

[0018] The display (160) may include, for example, a liquid crystal display (LCD), a light-emitting diode (LED) display, an organic light-emitting diode (OLED) display, a micro-electromechanical system (MEMS) display, or an electronic paper display. The display (160) may display various content (e.g., text, images, videos, icons, and / or symbols, etc.) to a user, for example. The display (160) may include a touch screen and may receive touch, gesture, proximity, or hovering input using, for example, an electronic pen or a part of the user's body. The communication interface (170) may establish communication between, for example, the electronic device (101) and an external device (e.g., a first external electronic device (102), a second external electronic device (104), or a server (106)). For example, the communication interface (170) can be connected to a network (162) via wireless or wired communication to communicate with an external device (e.g., a second external electronic device (104) or a server (106)).

[0019] Wireless communication may include cellular communication using at least one of, for example, LTE, LTE-A (LTE Advance), CDMA (code division multiple access), WCDMA (wideband CDMA), UMTS (universal mobile telecommunications system), WiBro (Wireless Broadband), or GSM (Global System for Mobile Communications). According to one embodiment, wireless communication may include at least one of, for example, WiFi (wireless fidelity), Bluetooth, Bluetooth Low Energy (BLE), Zigbee, NFC (near field communication), Magnetic Secure Transmission, Radio Frequency (RF), or Body Area Network (BAN). According to one embodiment, wireless communication may include GNSS. GNSS may be, for example, GPS (Global Positioning System), Glonass (Global Navigation Satellite System), Beidou Navigation Satellite System (hereinafter "Beidou"), or Galileo, the European global satellite-based navigation system. Hereinafter, in this document, "GPS" may be used interchangeably with "GNSS". Wired communication may include at least one of, for example, USB (universal serial bus), HDMI (high definition multimedia interface), RS-232 (recommended standard 232), power line communication, or POTS (plain old telephone service).The network (162) may include at least one of a telecommunications network, for example, a computer network (e.g., LAN or WAN), the Internet, or a telephone network.

[0020] Each of the first and second external electronic devices (102, 104) may be the same or a different type of device as the electronic device (101). According to various embodiments, all or part of the operations performed on the electronic device (101) may be performed on one or more other electronic devices (e.g., electronic devices (102, 104), or a server (106). According to one embodiment, when the electronic device (101) needs to perform a function or service automatically or upon request, the electronic device (101) may request at least some of the associated functions from another device (e.g., electronic devices (102, 104), or a server (106)) instead of performing the function or service itself or additionally. The other electronic device (e.g., electronic devices (102, 104), or a server (106)) may perform the requested function or additional functions and transmit the result to the electronic device (101). The electronic device (101) may provide the requested function or service by processing the received result as is or additionally. For this purpose, for example, cloud computing, distributed computing, or client-server computing technologies may be used.

[0022] FIG. 2 is a block diagram of an electronic device (201) according to various embodiments. The electronic device (201) may include, for example, all or part of the electronic device (101) shown in FIG. 1. The electronic device (201) may include one or more processors (e.g., AP) (210), a communication module (220), a subscriber identification module (224), a memory (230), a sensor module (240), an input device (250), a display (260), an interface (270), an audio module (280), a camera module (291), a power management module (295), a battery (296), an indicator (297), and a motor (298). The processor (210) may, for example, run an operating system or an application to control a number of hardware or software components connected to the processor (210) and perform various data processing and operations. The processor (210) may be implemented as, for example, a system on chip (SoC). According to one embodiment, the processor (210) may further include a graphics processing unit (GPU) and / or an image signal processor. The processor (210) may also include at least some of the components shown in FIG. 2 (e.g., a cellular module (221)). The processor (210) may process commands or data received from at least one of the other components (e.g., non-volatile memory) by loading them into volatile memory, and may store the resulting data in non-volatile memory.

[0023] The communication module (220) (e.g., communication interface (170)) may have the same or similar configuration. The communication module (220) may include, for example, a cellular module (221), a WiFi module (223), a Bluetooth module (225), a GNSS module (227), an NFC module (228), and an RF module (229). The cellular module (221) may, for example, provide voice calls, video calls, text services, or internet services through a communication network. According to one embodiment, the cellular module (221) may perform identification and authentication of an electronic device (201) within a communication network using a subscriber identification module (e.g., a SIM card) (224). According to one embodiment, the cellular module (221) may perform at least some of the functions that the processor (210) can provide. According to one embodiment, the cellular module (221) may include a communication processor (CP). According to some embodiments, at least some (e.g., two or more) of the cellular module (221), WiFi module (223), Bluetooth module (225), GNSS module (227), or NFC module (228) may be included in a single integrated chip (IC) or IC package. The RF module (229) may transmit and receive communication signals (e.g., RF signals), for example. The RF module (229) may include, for example, a transceiver, a power amp module (PAM), a frequency filter, a low noise amplifier (LNA), or an antenna. According to other embodiments, at least one of the cellular module (221), WiFi module (223), Bluetooth module (225), GNSS module (227), or NFC module (228) may transmit and receive RF signals through a separate RF module.The subscriber identification module (224) may include, for example, a card or an embedded SIM containing the subscriber identification module, and may include unique identification information (e.g., ICCID (integrated circuit card identifier)) or subscriber information (e.g., IMSI (international mobile subscriber identity)).

[0024] Memory (230) (e.g., memory (130)) may include, for example, internal memory (232) or external memory (234). Internal memory (232) may include, for example, at least one of volatile memory (e.g., DRAM, SRAM, or SDRAM), non-volatile memory (e.g., OTPROM (one time programmable ROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, flash memory, hard drive, or solid-state drive (SSD). External memory (234) may include a flash drive, for example, CF (compact flash), SD (secure digital), Micro-SD, Mini-SD, xD (extreme digital), MMC (multi-media card), or Memory Stick. External memory (234) may be functionally or physically connected to the electronic device (201) through various interfaces.

[0025] The sensor module (240) can, for example, measure physical quantities or detect the operating state of the electronic device (201) and convert the measured or detected information into an electrical signal. The sensor module (240) may include at least one of, for example, a gesture sensor (240A), a gyroscope sensor (240B), a barometric pressure sensor (240C), a magnetic sensor (240D), an accelerometer sensor (240E), a grip sensor (240F), a proximity sensor (240G), a color sensor (240H) (e.g., an RGB (red, green, blue) sensor), a biosensor (240I), a temperature / humidity sensor (240J), an illuminance sensor (240K), or a UV (ultra violet) sensor (240M). Additionally or generally, the sensor module (240) may include, for example, an e-nose sensor, an electromyography (EMG) sensor, an electro-encyclogram (EEG) sensor, an electrocardiogram (ECG) sensor, an infrared (IR) sensor, an iris sensor, and / or a fingerprint sensor. The sensor module (240) may further include a control circuit for controlling at least one of the sensors included therein. In some embodiments, the electronic device (201) may further include a processor configured to control the sensor module (240) as part of or separately from the processor (210), so as to control the sensor module (240) while the processor (210) is in a sleep state.

[0026] The input device (250) may include, for example, a touch panel (252), a (digital) pen sensor (254), a key (256), or an ultrasonic input device (258). The touch panel (252) may use at least one of, for example, capacitive, resistive, infrared, or ultrasonic methods. Additionally, the touch panel (252) may further include a control circuit. The touch panel (252) may further include a tactile layer to provide a tactile response to the user. The (digital) pen sensor (254) may, for example, be part of the touch panel or include a separate recognition sheet. The key (256) may include, for example, a hardware button, an optical key, or a keypad. The ultrasonic input device (258) can detect ultrasonic waves generated from the input tool through a microphone (e.g., microphone (288)) and verify data corresponding to the detected ultrasonic waves.

[0027] A display (260) (e.g., display (160)) may include a panel (262), a holographic device (264), a projector (266), and / or a control circuit for controlling these. The panel (262) may be implemented, for example, in a flexible, transparent, or wearable manner. The panel (262) may be composed of a touch panel (252) and one or more modules. According to one embodiment, the panel (262) may include a pressure sensor (or force sensor) capable of measuring the intensity of pressure for a user's touch. The pressure sensor may be implemented integrally with the touch panel (252) or as one or more sensors separate from the touch panel (252). The holographic device (264) may display a three-dimensional image in mid-air using light interference. The projector (266) may display an image by projecting light onto a screen. The screen may be located, for example, inside or outside the electronic device (201). The interface (270) may include, for example, HDMI (272), USB (274), optical interface (276), or D-sub (D-subminiature) (278). The interface (270) may include, for example, the communication interface (170) shown in FIG. 1. Additionally or alternatively, the interface (270) may include, for example, a mobile high-definition link (MHL) interface, an SD card / multi-media card (MMC) interface, or an infrared data association (IrDA) standard interface.

[0028] The audio module (280) can, for example, convert sound and electrical signals bidirectionally. At least some components of the audio module (280) may be included in, for example, the input / output interface (145) shown in FIG. 1. The audio module (280) may process sound information input or output through, for example, a speaker (282), a receiver (284), earphones (286), or a microphone (288), etc. The camera module (291) is, for example, a device capable of capturing still images and video, and according to one embodiment, may include one or more image sensors (e.g., a front sensor or a rear sensor), a lens, an image signal processor (ISP), or a flash (e.g., an LED or a xenon lamp, etc.). The power management module (295) may, for example, manage the power of the electronic device (201). According to one embodiment, the power management module (295) may include a power management integrated circuit (PMIC), a charging IC, or a battery or fuel gauge. The PMIC may have wired and / or wireless charging methods. Wireless charging methods include, for example, magnetic resonance methods, magnetic induction methods, or electromagnetic wave methods, and may further include additional circuits for wireless charging, for example, coil loops, resonant circuits, or rectifiers. A battery gauge may measure, for example, the remaining charge of the battery (296), voltage, current, or temperature during charging. The battery (296) may include, for example, a rechargeable battery and / or a solar cell.

[0029] The indicator (297) can display a specific state of the electronic device (201) or a part thereof (e.g., processor (210)), for example, a boot state, a message state, or a charging state. The motor (298) can convert an electrical signal into a mechanical vibration and can generate vibration or haptic effects. The electronic device (201) is, for example, DMB (digital multimedia broadcasting), DVB (digital video broadcasting), or mediaFlo TM It may include a mobile TV support device (e.g., GPU) capable of processing media data according to standards such as ). Each of the components described in this document may consist of one or more components, and the name of the component may vary depending on the type of electronic device. In various embodiments, the electronic device (e.g., electronic device (201)) may have some components omitted, may include additional components, or may be composed of some of the components combined into a single entity, which can perform the same function as the components prior to combination.

[0030] In various embodiments of the present invention, the electronic device (201) (or electronic device (101)) may include a housing comprising a front, a rear, and a side that surrounds the space between the front and the rear. A touchscreen display (e.g., display (260)) is placed inside the housing and may be exposed through the front. A microphone (288) is placed inside the housing and may be exposed through a portion of the housing. At least one speaker (282) is placed inside the housing and may be exposed through another portion of the housing. A hardware button (e.g., key (256)) may be placed in another portion of the housing or configured to be displayed on the touchscreen display. A wireless communication circuit (e.g., communication module (220)) may be located inside the housing. The processor (210) (or processor (120)) is located within the housing and may be electrically connected to the touchscreen display, the microphone (288), the speaker (282), and the wireless communication circuit. The memory (230) (or memory (130)) is located within the housing and may be electrically connected to the processor (210).

[0031] In various embodiments of the present invention, the memory (230) is configured to store a first application program including a first user interface for receiving text input, and the memory (230) stores instructions that, at execution, cause the processor (210) to perform a first operation and a second operation, the first operation is to receive a first type of user input through the button while the first user interface is not displayed on the touchscreen display, and after receiving the first type of user input, to receive a first user utterance through the microphone (288), provide first data regarding the first user utterance to an external server including an automatic speech recognition (ASR) and intelligence system, and after providing the first data, to receive at least one command from the external server to perform a task generated by the intelligence system in response to the first user utterance, and the second operation is to perform the first on the touchscreen display 1. While the user interface is being displayed, the first user input is received through the button, and after receiving the first type of user input, the second user utterance is received through the microphone (288), and second data regarding the second user utterance is provided to the external server, and after providing the second data, data regarding text generated by automatic speech recognition from the second user utterance is received from the server, but the command generated by the intelligent system is not received, and the text can be entered into the first user interface.

[0032] In various embodiments of the present invention, the button may include a physical key located on the side of the housing.

[0033] In various embodiments of the present invention, the first type of user input may be one of a single press of the button, two presses of the button, three presses of the button, holding down after one press of the button, or two presses of the button and holding down.

[0034] In various embodiments of the present invention, the instructions may further cause the processor to display the first user interface along with a virtual keyboard. The button may not be part of the virtual keyboard.

[0035] In various embodiments of the present invention, the instructions may further cause the processor (210) to receive data for text generated by ASR from the first user utterance in the first operation from the external server.

[0036] In various embodiments of the present invention, the first application program may include at least one of a note application program, an email application program, a web browser application program, or a calendar application program.

[0037] In various embodiments of the present invention, the first application program includes a message application, and the instructions may further cause the processor (210) to automatically transmit the input text through the wireless communication circuit when a selected time period has elapsed after the text has been input.

[0038] In various embodiments of the present invention, the instructions further cause the processor (210) to perform a third operation, the third operation being to receive a second type of user input through the button while displaying the first user interface on the touchscreen display, receive a third user utterance through the microphone after receiving the second type of user input, provide third data regarding the third user utterance to the external server, and after providing the third data, receive at least one command from the external server to perform a task generated by the intelligent system in response to the third user utterance.

[0039] In various embodiments of the present invention, the instructions further cause the processor (210) to perform a fourth operation, the fourth operation may receive a second type of user input through the button while the first user interface is not displayed on the touchscreen display, receive a fourth user utterance through the microphone (288) after receiving the second type of user input, provide fourth data for the fourth user utterance to the external server, receive at least one command from the external server to perform a task generated by the intelligent system in response to the fourth user utterance after providing the fourth data, receive a fifth user utterance through the microphone, provide fifth data for the fifth user utterance to the external server, and receive at least one command from the external server to perform a task generated by the intelligent system in response to the fifth user utterance after providing the fifth data.

[0040] In various embodiments of the present invention, the first type of user input and the second type of user input are different from each other and may be selected from one of a single press of the button, a double press of the button, a triple press of the button, a press held after a single press of the button, or a double press and a press held of the button.

[0041] In various embodiments of the present invention, the memory (230) is further configured to store a second application program including a second user interface for receiving text input, and the instructions further cause the processor (210) to perform a third operation when executed, the third operation is to receive the first type of user input through the button while displaying the second user interface, receive a third user utterance through the microphone after the first type of user input is received, provide third data regarding the third user utterance to the external server, and after providing the third data, receive data regarding text generated by ASR from the third user utterance from the external server, without receiving commands generated by the intelligent system, input the text into the second user interface, and when the selected time period is exceeded after inputting the text, the input text can be automatically transmitted through the wireless communication circuit.

[0042] In various embodiments of the present invention, the memory (230) is configured to store a first application program including a first user interface for receiving text input, and the memory (230) stores instructions that, upon execution, cause the processor (210) to perform a first operation and a second operation, the first operation is to receive a first type of user input through the button, and after receiving the first type of user input, to receive a first user utterance through the microphone (288), to provide first data regarding the first user utterance to an external server including an automatic speech recognition (ASR) and an intelligence system, and after providing the first data, to receive at least one command from the external server for performing a task generated by the intelligence system in response to the first user utterance, and the second operation is to receive a second type of user input through the button, and after receiving the second type of user input, A second user utterance is received through the microphone (288), and second data regarding the second user utterance is provided to the external server. After providing the second data, data regarding text generated by ASR from the second user utterance is received from the server, while commands generated by the intelligent system are not received, and the text can be entered into the first user interface.

[0043] In various embodiments of the present invention, the instructions may further cause the processor (210) to display the first user interface together with a virtual keyboard, and the buttons may not be part of the virtual keyboard.

[0044] In various embodiments of the present invention, the instructions may further cause the processor (210) to receive data for text generated by the ASR from the first user utterance within the first operation from the external server.

[0045] In various embodiments of the present invention, the first application program may include at least one of a note application program, an email application program, a web browser application program, or a calendar application program.

[0046] In various embodiments of the present invention, the first application program includes a message application, and the instructions may further cause the processor (210) to automatically transmit the input text through the wireless communication circuit when a selected time period has elapsed after the text has been input.

[0047] In various embodiments of the present invention, the instructions may further cause the processor (210) to perform the first operation independently of the display on the display of the first user interface.

[0048] In various embodiments of the present invention, the instructions may further cause the processor (210) to perform the second operation when at least one of the electronic device is locked or the touchscreen display is turned off.

[0049] In various embodiments of the present invention, the instructions may further cause the processor (210) to perform the second operation while displaying the first user interface on the touchscreen display.

[0050] In various embodiments of the present invention, the memory (230) may store an instruction that, at execution, causes the processor (210) to receive a user utterance through the microphone (288) and to perform at least one of automatic speech recognition (ASR) or natural language understanding (NLU) to transmit information related to whether to perform natural language understanding on the text obtained by performing ASR on the data for the user utterance, along with data for the user utterance, and if the information indicates that natural language understanding will not be performed, to receive the text for the data for the user utterance from the external server, and if the information indicates that natural language understanding will be performed, to receive a command obtained as a result of performing natural language understanding on the text from the external server.

[0052] FIG. 3 is a block diagram of a program module according to various embodiments. According to one embodiment, the program module (310) (e.g., program (140)) may include an operating system that controls resources related to an electronic device (e.g., electronic device (101)) and / or various applications running on the operating system (e.g., application programs (147)). The operating system is, for example, Android TM , iOS TM , Windows TM , Symbian TM , Tizen TM , or Bada TMIt may include. Referring to FIG. 3, the program module (310) may include a kernel (320) (e.g., kernel (141)), middleware (330) (e.g., middleware (143)), API (360) (e.g., API (145)), and / or an application (370) (e.g., application program (147)). At least a portion of the program module (310) may be preloaded onto an electronic device or downloaded from an external electronic device (e.g., electronic device (102, 104), server (106), etc.).

[0053] The kernel (320) may include, for example, a system resource manager (321) and / or a device driver (323). The system resource manager (321) may perform control, allocation, or reclamation of system resources. According to one embodiment, the system resource manager (321) may include a process management unit, a memory management unit, or a file system management unit. The device driver (323) may include, for example, a display driver, a camera driver, a Bluetooth driver, a shared memory driver, a USB driver, a keypad driver, a WiFi driver, an audio driver, or an IPC (inter-process communication) driver. The middleware (330) may, for example, provide functions commonly required by the application (370) or provide various functions to the application (370) via an API (360) so that the application (370) can use limited system resources within the electronic device. According to one embodiment, the middleware (330) may include at least one of a runtime library (335), an application manager (341), a window manager (342), a multimedia manager (343), a resource manager (344), a power manager (345), a database manager (346), a package manager (347), a connectivity manager (348), a notification manager (349), a location manager (350), a graphics manager (351), or a security manager (352).

[0054] The runtime library (335) may include, for example, library modules used by the compiler to add new functions through a programming language while the application (370) is running. The runtime library (335) may perform input / output management, memory management, or arithmetic function processing. The application manager (341) may, for example, manage the lifecycle of the application (370). The window manager (342) may manage GUI resources used on the screen. The multimedia manager (343) may identify the format required for the playback of media files and perform encoding or decoding of the media files using a codec that matches the format. The resource manager (344) may manage the source code or memory space of the application (370). The power manager (345) may, for example, manage the capacity or power of the battery and provide power information required for the operation of the electronic device. According to one embodiment, the power manager (345) may be linked with the BIOS (basic input / output system). The database manager (346) can, for example, create, search, or change the database to be used in the application (370). The package manager (347) can manage the installation or update of the application distributed in the form of a package file.

[0055] The connectivity manager (348) can manage wireless connections, for example. The notification manager (349) can provide events to the user, such as arrival messages, appointments, and proximity notifications, for example. The location manager (350) can manage location information of the electronic device, for example. The graphics manager (351) can manage graphic effects or related user interfaces to be provided to the user, for example. The security manager (352) can provide system security or user authentication, for example. According to one embodiment, the middleware (330) may include a telephony manager for managing voice or video call functions of the electronic device or a middleware module capable of forming a combination of the functions of the aforementioned components. According to one embodiment, the middleware (330) may provide modules specialized for each type of operating system. The middleware (330) may dynamically delete some existing components or add new components. The API (360) is, for example, a set of API programming functions and may be provided in different configurations depending on the operating system. For example, in the case of Android or iOS, one set of APIs may be provided per platform, and in the case of Tizen, two or more sets of APIs may be provided per platform.

[0056] The application (370) may include, for example, a home (371), a dialer (372), an SMS / MMS (373), an IM (instant message) (374), a browser (375), a camera (376), an alarm (377), a contact (378), a voice dialer (379), an email (380), a calendar (381), a media player (382), an album (383), a watch (384), a healthcare application (e.g., measuring exercise volume or blood sugar, etc.), or an application for providing environmental information (e.g., atmospheric pressure, humidity, or temperature information). According to one embodiment, the application (370) may include an information exchange application capable of supporting information exchange between an electronic device and an external electronic device. The information exchange application may include, for example, a notification relay application for transmitting specific information to an external electronic device, or a device management application for managing an external electronic device. For example, a notification delivery application may transmit notification information generated by another application of the electronic device to an external electronic device, or receive notification information from an external electronic device and provide it to the user. A device management application may install, delete, or update functions of an external electronic device communicating with the electronic device (e.g., turning on / off of the external electronic device itself (or some components) or adjusting the brightness (or resolution) of a display), or applications running on the external electronic device. According to one embodiment, the application (370) may include an application specified according to the attributes of the external electronic device (e.g., a health management application for a mobile medical device). According to one embodiment, the application (370) may include an application received from an external electronic device.At least a portion of the program module (310) may be implemented (e.g., executed) by software, firmware, hardware (e.g., processor (210)), or a combination of at least two of these, and may include a module, program, routine, instruction set, or process for performing one or more functions.

[0058] FIG. 4 illustrates a learning process for learning a meta-learner according to various embodiments.

[0059] The meta-learning algorithm of Fig. 4 uses knowledge shared about changing tasks as the parameters of the neural network ( t Learning is performed in the form of ). In the present invention, a processor (120) combined with a neural network can be optimized immediately with only a small number of samples of a new task, and the signal processing module optimized in this way receives the signal of the neural network as input and performs 'task-adaptive processing' optimized for the current task, thereby enabling greatly improved inference.

[0060] The present invention relates to a combined system of a neural network and a processor (120) that receives a signal from the neural network as input. Specifically, the output signal of the output layer of the neural network, a hidden layer signal, or a feature is received as input to the processor (120) and processed.

[0061] The inference of a neural network that is lacking is greatly improved through additional processing by the processor (120) of the present invention. The present invention is not limited to the structure and form of the neural network and contains a specific design strategy for the processor (120). Based on this combined system and learning algorithm, it is possible to achieve greatly improved inference performance in learning situations that are difficult to handle with neural network learning alone.

[0063] FIG. 5 illustrates configurations for providing an artificial intelligence model included in a server (e.g., the electronic device (101) of FIG. 1) according to various embodiments.

[0064] FIG. 6 shows an embodiment in which a data collection unit operates according to various embodiments.

[0065] FIG. 7 shows an embodiment in which a data preprocessing unit operates according to various embodiments.

[0066] FIG. 8 shows an embodiment in which a data visualization unit operates according to various embodiments.

[0067] FIG. 9 shows an embodiment in which a model verification unit operates according to various embodiments.

[0068] According to various embodiments, the server (101) may include a data collection unit (501), a data preprocessing unit (502), a data visualization unit (503), a model training unit (504), a model verification unit (505), a model provision unit (506), and an HMI (Human-Machine Interface) unit (507).

[0069] According to one embodiment, the data collection unit (501) can collect data from a plurality of servers using web crawling and convert the collected data into numerical data. For example, referring to FIG. 5, the data collection unit (501) can collect data from a plurality of servers that process various data (e.g., public administration, environmental weather, industrial employment, transportation and logistics, etc.) by accessing the websites managed by the servers. At this time, the collected data may have a format such as an image, video, voice, text, or number, and the data collection unit (501) can convert the data obtained by crawling into a numerical (number) data format.

[0070] According to one embodiment, the data preprocessing unit (502) can receive numeric data from the data collection unit (501), preprocess the numeric data to normalize the data, and generate training data by performing a dimension transformation on the normalized data. For example, referring to FIG. 7, the data preprocessing unit (502) can generate normalized data by removing noise, processing missing values, combining column data, and performing unit unification as a preprocessing process for numeric data, and can generate training data by performing a dimension transformation so that the normalized data can be commonly used in a meta-learning algorithm model.

[0071] According to one embodiment, the data visualization unit (503) can visualize training data received from the data preprocessing unit (502). For example, referring to FIG. 8, the data visualization unit (503) can generate a UI (user interface) that allows various visualization information to be easily displayed to the user by utilizing the data before and after preprocessing. Specifically, the training data can generate a template applicable to various data and representation methods (e.g., area, bar, bubble, chart, etc.) to generate a UI that can be viewed by a model user corresponding to the user. According to one embodiment, the data visualization unit (503) may include a data visualization tool that can be easily used by a person skilled in the art.

[0072] According to one embodiment, the model training unit (504) can train an artificial intelligence model using training data according to a meta-learning method. For meta-learning model training, the following three approaches may be supported.

[0073] - Method for training the Metric-based Efficient Distance Metric

[0074] - Model-based RNN method using External / Internal Memory

[0075] - Model parameter optimization method for optimization-based fast learning

[0076] The model training unit (504) can train the following artificial intelligence model according to the meta-learning methods described above.

[0077] According to one embodiment, when the model training unit (504) receives training data corresponding to image data for image classification, it can train an artificial intelligence model based on a Convolutional Neural Network (CNN).

[0078] The model training unit (504) can train a CNN-based artificial intelligence model when it receives training data corresponding to image data for object detection.

[0079] When the model training unit (504) receives training data corresponding to table data for regression / classification, it can train an artificial intelligence model based on Extreme Gradient Boosting (XGBoost).

[0080] When the model training unit (504) receives training data corresponding to table data for prediction, it can train an artificial intelligence model based on Long Short-term Memory (LSTM).

[0081] When the model training unit (504) receives training data corresponding to text data for text classification, it can train an artificial intelligence model based on a Recurrent Neural Network (RNN).

[0082] The model training unit (504) can train a CNN-based artificial intelligence model when it receives training data corresponding to text data for text item extraction.

[0083] The model training unit (504) can train an artificial intelligence model based on one of CNN, RNN, and Generative Adversarial Networks (GAN) when receiving training data corresponding to video data for motion recognition, image classification, and object tracking.

[0084] According to one embodiment, the model verification unit (505) can verify the reliability of the artificial intelligence model by inputting verification / evaluation data corresponding to the training data into the artificial intelligence model. For example, referring to FIG. 9, the model verification unit (505) can verify the reliability regarding whether the artificial intelligence model is suitable for providing to other users by inputting verification / evaluation data into the artificial intelligence model trained based on the training data. According to one embodiment, the model verification algorithm may use an artificial intelligence model verification algorithm that is obvious to those skilled in the art, and if the reliability of the artificial intelligence model is below a threshold value pre-set by the administrator of the server (101) as a result of the verification of the artificial intelligence model, the fact that the reliability of the artificial intelligence model has failed can be transmitted to the model training unit (504), and in this case, the model training unit (504) can retrain the artificial intelligence model. Meanwhile, if the reliability is above the threshold value, the fact that the reliability of the artificial intelligence model has passed can be transmitted to the model provision (506).

[0085] According to one embodiment, the model providing unit (506) may provide an API (Application Programming Interface) regarding the artificial intelligence model so that multiple users can access the artificial intelligence model when the reliability of the artificial intelligence model is above a threshold value.

[0086] According to one embodiment, the Human-Machine Interface (HMI) unit (507) may provide a graphical user interface that allows the plurality of users to connect to the server and perform a plurality of operations related to the artificial intelligence model. The artificial intelligence model may be implemented through the use of a software-based model design tool. Such a motion design tool may utilize a graphical user interface (GUI) to present options for defining the input data, output data, and other characteristics of the model to the artificial intelligence model designer, and may be implemented as a series of one or more web pages and / or web-based applications. Server users may use the model design tool to perform various tasks in an organized and efficient manner. In particular, the model design tool may be a so-called "low-code / no-code" solution, and with this solution, designers may write very little program code or no code at all to implement the artificial intelligence model. For example, the HMI section (507) can provide a GUI for selecting data stored on a server, a GUI for uploading data to the server that the user intends to learn, and a GUI for performing specific tasks (prediction, tracking, classification, etc.) using the data.

[0088] According to various embodiments, a server providing an artificial intelligence model trained according to a meta-learning method may include: a data collection unit (501) that collects data from multiple servers using web crawling and converts the collected data into numerical data; a data preprocessing unit (502) that generates training data by preprocessing the numerical data to normalize the data and performing a dimension transformation on the normalized data; a data visualization unit (503) that visualizes the training data; a model training unit (504) that trains an artificial intelligence model using the training data according to a meta-learning method; a model verification unit (505) that verifies the reliability of the artificial intelligence model by inputting verification / evaluation data corresponding to the training data into the artificial intelligence model; a model providing unit (506) that allows multiple users to access the artificial intelligence model when the reliability of the artificial intelligence model is above a threshold value; and a Human-Machine Interface (HMI) unit (507) that provides a graphical user interface that allows the multiple users to connect to the server and perform multiple operations related to the artificial intelligence model.

[0089] According to various embodiments, the model training unit may train a Convolutional Neural Network (CNN)-based artificial intelligence model when receiving training data corresponding to image data for image classification, train a CNN-based artificial intelligence model when receiving training data corresponding to image data for object detection, train an Extreme Gradient Boosting (XGBoost)-based artificial intelligence model when receiving training data corresponding to table data for regression / classification, train a Long Short-term Memory (LSTM)-based artificial intelligence model when receiving training data corresponding to table data for prediction, train a Recurrent Neural Network (RNN)-based artificial intelligence model when receiving training data corresponding to text data for text classification, train a CNN-based artificial intelligence model when receiving training data corresponding to text data for text item extraction, and train an artificial intelligence model based on one of CNN, RNN, and Generative Adversarial Networks (GAN) when receiving training data corresponding to video data for motion recognition, video classification, and object tracking. there is.

[0090] According to various embodiments, the model training unit may receive the fact of reliability failure from the model verification unit and retrain the artificial intelligence model when the reliability is below the threshold value as a result of verifying the artificial intelligence model.

[0092] As used in this document, the terms “module” or “part” include a unit composed of hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit, for example. “Module” or “part” may be a component formed integrally or a minimum unit or part thereof that performs one or more functions. “Module” or “part” may be implemented mechanically or electronically and may include, for example, an application-specific integrated circuit (ASIC) chip, field-programmable gate arrays (FPGAs), or programmable logic device known or to be developed that performs certain operations, and may be executed by a processor (120). At least part of the device (e.g., modules or functions thereof) or method (e.g., operations) according to various embodiments may be implemented as instructions stored in a computer-readable storage medium (e.g., memory (130)) in the form of a program module. When the above instruction is executed by a processor (e.g., processor (120)), the processor may perform a function corresponding to the above instruction. Computer-readable recording media may include a hard disk, a floppy disk, a magnetic medium (e.g., magnetic tape), an optical recording medium (e.g., CD-ROM, DVD, magneto-optical medium (e.g., floptical disk), built-in memory, etc. Instructions may include code generated by a compiler or code that can be executed by an interpreter. A module or program module according to various embodiments may include at least one of the aforementioned components, some of which may be omitted, or additionally include other components. Operations performed by a module, program module, or other components according to various embodiments may be executed sequentially, in parallel, iteratively, or heuristically, or at least some operations may be executed in a different order, omitted, or other operations may be added.

[0093] Furthermore, the embodiments disclosed in this document are presented for the purpose of explaining and understanding the disclosed technical content and are not intended to limit the scope of this disclosure. Accordingly, the scope of this disclosure should be interpreted to include all modifications or various other embodiments based on the technical concept of this disclosure.

Claims

Claim 1 A server providing an artificial intelligence model trained according to a meta-learning method comprises: a data collection unit (501) that collects data from multiple servers using web crawling and converts the collected data into numerical data; a data preprocessing unit (502) that generates training data by preprocessing the numerical data to normalize the data and performing a dimension transformation on the normalized data; a data visualization unit (503) that visualizes the training data; a model training unit (504) that trains an artificial intelligence model using the training data according to a meta-learning method; a model verification unit (505) that verifies the reliability of the artificial intelligence model by inputting verification / evaluation data corresponding to the training data into the artificial intelligence model; a model providing unit (506) that enables multiple users to access the artificial intelligence model when the reliability of the artificial intelligence model is above a threshold value; and an HMI (Human-Machine Interface) unit (507) that provides a graphical user interface that enables multiple users to connect to the server and perform multiple operations related to the artificial intelligence model. The model training unit receives a fact of reliability failure from the model verification unit when the reliability of the artificial intelligence model is below the threshold value as a result of verification. A server providing an artificial intelligence model that retrains the above artificial intelligence model. Claim 2 In claim 1, the model training unit may train a Convolutional Neural Network (CNN)-based artificial intelligence model when receiving training data corresponding to image data for image classification, the model training unit may train a CNN-based artificial intelligence model when receiving training data corresponding to image data for object detection, the model training unit may train an Extreme Gradient Boosting (XGBoost)-based artificial intelligence model when receiving training data corresponding to table data for regression / classification, the model training unit may train a Long Short-term Memory (LSTM)-based artificial intelligence model when receiving training data corresponding to table data for prediction, the model training unit may train a Recurrent Neural Network (RNN)-based artificial intelligence model when receiving training data corresponding to text data for text classification, the model training unit may train a CNN-based artificial intelligence model when receiving training data corresponding to text data for text item extraction, and the model training unit may train an artificial intelligence model based on one of CNN, RNN, and Generative Adversarial Networks (GAN) when receiving training data corresponding to video data for motion recognition, video classification, and object tracking. A server that provides an artificial intelligence model for training. Claim 3 delete

Images