Search

KR-20260064477-A - SERVER FOR RECOMMENDING CUSTOMIZED MODEL BASED ON PREDICTIVE MODEL AND METHOD OPERATION THEREOF

KR20260064477AKR 20260064477 AKR20260064477 AKR 20260064477AKR-20260064477-A

Abstract

According to various embodiments, a server that recommends a user-customized model based on a prediction model comprises: a communication interface; and a processor; wherein the processor receives at least one user data from an external electronic device of the user through the communication interface, inputs the at least one user data into a first sub-deep learning model to determine text data for the user, inputs the at least one user data into a second sub-deep learning model to determine image data for the user, inputs at least one of the text data and/or the image data into the prediction model to determine a customized model for the user, and is configured to transmit the customized model to the external electronic device of the user through the communication interface; and wherein the prediction model is learned based on a plurality of text data and a plurality of image data, a plurality of customized models, a first result data in which the plurality of customized models are determined to be accurate, a second result data in which the plurality of customized models are determined to have some errors, and a third result data in which the plurality of customized models are determined to have errors.

Inventors

  • 도영태
  • 최수희

Assignees

  • 주식회사 누리에에이아이

Dates

Publication Date
20260507
Application Date
20250716
Priority Date
20241030

Claims (18)

  1. In a server that recommends user-customized models based on a predictive model, Communication interface; Includes a processor; and The above processor is, Through the above communication interface, at least one user data is received from the user's external electronic device, and The above at least one user data is input into a first sub-deep learning model to determine text data for the user, and The above at least one user data is input into a second sub-deep learning model to determine image data for the user, and At least one of the text data and/or image data is input into the prediction model to determine a customized model for the user, and The customized model is configured to be transmitted to the user's external electronic device through the above communication interface, and The above prediction model is, Learning based on a plurality of text data and a plurality of image data, a plurality of customized models, a first result data in which the plurality of customized models are determined to be accurate, a second result data in which the plurality of customized models are determined to have some errors, and a third result data in which the plurality of customized models are determined to have errors. Server.
  2. In Article 1, The above at least one user data is, Sentence data written by the user, translation data written by the user, summary data written by the user, conversational data written by the user, voice data detected or emitted by the user, tactile data detected or emitted by the user, olfactory data detected or emitted by the user, input image data verified by the user's eyes or entered by the user, and analysis data of the user, mathematical operation data of the user, Server.
  3. In Article 2, The above processor is, At least one of the sentence data, the translation data, the summary data, and the conversational data is input into the first sub-deep learning model to determine the text data, and If it is determined that the above text data does not exceed a preset text threshold, the above text data is classified as first text result data, and If it is determined that the above text data exceeds the above text threshold, the text data is set to be classified as second text result data, and The above-mentioned first sub-deep learning model is, Learning based on multiple user data, multiple sentence data, multiple translation data, multiple summary data, and multiple conversational data, and multiple text data, multiple first text result data, and multiple second text result data, Server.
  4. In Paragraph 3, The above processor is, At least one of the above input image data, the above analysis data, and the above at least one user data is input into the above second sub-deep learning model to determine the image data, and If it is determined that the above image data does not exceed a preset image threshold, the above image data is classified as first image result data, and If it is determined that the above image data exceeds the above image threshold, the above image data is set to be classified as second image result data, and The above second sub-deep learning model is, Learning based on multiple user data, multiple analysis data, and multiple input image data, multiple image data, multiple first image result data, and multiple second image result data, Server.
  5. In Paragraph 4, The above processor is, At least one of the above at least one user data, the voice data, the tactile data, and the olfactory data is input into a third sub-deep learning model to determine additional data for the user, and If it is determined that the above-mentioned extra data does not exceed a preset extra threshold, the above-mentioned extra data is classified as first extra result data, and If it is determined that the above-mentioned extra data exceeds the above-mentioned extra threshold, the above-mentioned extra data is configured to be classified as second extra result data, and The above third sub-deep learning model is, Learning based on multiple user data, multiple voice data, multiple tactile data, and multiple olfactory data, and multiple redundant data, multiple first redundant result data, and multiple second redundant result data, Server.
  6. In Article 5, The above processor is, The above text data is input into the prediction model based on [Mathematical Formula 1] below to generate a first quality loss function, a first cost loss function, a first efficiency loss function, and a first total loss function for the text data, respectively, and The above image data is input into the prediction model based on [Mathematical Formula 1] below to generate a second quality loss function, a second cost loss function, a second efficiency loss function, and a second total loss function for the image data, respectively, and The above redundant data is input into the above prediction model based on [Mathematical Formula 1] below, configured to generate a third quality loss function, a third cost loss function, a third efficiency loss function, and a third total loss function for the above redundant data, respectively. Server. [Mathematical Formula 1] (1) (2) (3) (4) Here, is the first total loss function, the second total loss function, and the third total loss function, and is the first quality loss function, the second quality loss function, and the third quality loss function, and is the first cost loss function, the second cost loss function, and the third cost loss function, and is the first efficiency loss function, the second efficiency loss function, and the third efficiency loss function, and is the respective weight for the quality loss function, cost loss function, and efficiency loss function, and is the value obtained by splitting the correct answer score for the domain, and is the completeness score value for the domain, and is a weight value for calculating the above quality loss function, and is the calculated value for the token relative to the price, and is the sum of other processing costs excluding the price, and is a weight value for calculating the above cost-loss function, and is a ratio value to the actual processing time, and is a ratio value to the number of tokens used, and is a weight value for calculating the above efficiency loss function.
  7. In Article 6, The above prediction model is, A plurality of redundant data, a plurality of first text result data, a plurality of second text result data, a plurality of first image result data, a plurality of second image result data, a plurality of first redundant result data, and a plurality of second redundant result data, and additionally trained based on a plurality of total loss functions, a plurality of quality loss functions, a plurality of cost loss functions, and a plurality of efficiency loss functions, Server.
  8. In Article 6, The above processor is, If the above customized model is below a preset first model threshold, it is determined that the above customized model is a user-customized model, and If the customized model is greater than or equal to the first threshold value or less than or equal to the second threshold value which is set to be greater than the first threshold value, the customized model is configured to determine that some error has occurred. Server.
  9. In Article 8, The above processor is, If the customized model is greater than or equal to the first threshold and the second threshold, it is determined that the error of the customized model is very serious, and the prediction model is retrained only if the customized model is greater than or equal to the first threshold and the second threshold, and Through the communication interface above, the customized model is configured to be transmitted to the user's external electronic device, limited to the customized model that does not exceed the second threshold value. Server.
  10. In a method for operating a server that recommends a user-customized model based on a prediction model, The above method is, Through a communication interface, at least one user data is received from an external electronic device of the user, and Through a processor, at least one user data is input into a first sub-deep learning model to determine text data for the user, and Through the above processor, at least one user data is input into a second sub-deep learning model to determine image data for the user, and Through the above processor, at least one of the text data and/or image data is input into the prediction model to determine a customized model for the user, and The custom model is configured to be transmitted to the user's external electronic device through the above communication interface, and The above prediction model is, Learning based on a plurality of text data and a plurality of image data, a plurality of customized models, a first result data in which the plurality of customized models are determined to be accurate, a second result data in which the plurality of customized models are determined to have some errors, and a third result data in which the plurality of customized models are determined to have errors. method.
  11. In Article 10, The above at least one user data is, Sentence data written by the user, translation data written by the user, summary data written by the user, conversational data written by the user, voice data detected or emitted by the user, tactile data detected or emitted by the user, olfactory data detected or emitted by the user, input image data verified by the user's eyes or entered by the user, and analysis data of the user, mathematical operation data of the user, method.
  12. In Article 11, The above method is, At least one of the sentence data, the translation data, the summary data, and the conversational data is input into the first sub-deep learning model to determine the text data, and If it is determined that the above text data does not exceed a preset text threshold, the above text data is classified as first text result data, and If it is determined that the above text data exceeds the above text threshold, the text data is set to be classified as second text result data, and The above-mentioned first sub-deep learning model is, Learning based on multiple user data, multiple sentence data, multiple translation data, multiple summary data, and multiple conversational data, and multiple text data, multiple first text result data, and multiple second text result data, method.
  13. In Article 12, The above method is, At least one of the above input image data, the above analysis data, and the above at least one user data is input into the above second sub-deep learning model to determine the image data, and If it is determined that the above image data does not exceed a preset image threshold, the above image data is classified as first image result data, and If it is determined that the above image data exceeds the above image threshold, the above image data is set to be classified as second image result data, and The above second sub-deep learning model is, Learning based on multiple user data, multiple analysis data, and multiple input image data, multiple image data, multiple first image result data, and multiple second image result data, method.
  14. In Article 13, The above method is, At least one of the above at least one user data, the voice data, the tactile data, and the olfactory data is input into a third sub-deep learning model to determine additional data for the user, and If it is determined that the above-mentioned extra data does not exceed a preset extra threshold, the above-mentioned extra data is classified as first extra result data, and If it is determined that the above-mentioned extra data exceeds the above-mentioned extra threshold, the above-mentioned extra data is configured to be classified as second extra result data, and The above third sub-deep learning model is, Learning based on multiple user data, multiple voice data, multiple tactile data, and multiple olfactory data, and multiple redundant data, multiple first redundant result data, and multiple second redundant result data, method.
  15. In Article 14, The above method is, The above text data is input into the prediction model based on [Mathematical Formula 1] below to generate a first quality loss function, a first cost loss function, a first efficiency loss function, and a first total loss function for the text data, respectively, and The above image data is input into the prediction model based on [Mathematical Formula 1] below to generate a second quality loss function, a second cost loss function, a second efficiency loss function, and a second total loss function for the image data, respectively, and The above redundant data is input into the above prediction model based on [Mathematical Formula 1] below, configured to generate a third quality loss function, a third cost loss function, a third efficiency loss function, and a third total loss function for the above redundant data, respectively. method. [Mathematical Formula 1] (1) (2) (3) (4) Here, is the first total loss function, the second total loss function, and the third total loss function, and is the first quality loss function, the second quality loss function, and the third quality loss function, and is the first cost loss function, the second cost loss function, and the third cost loss function, and is the first efficiency loss function, the second efficiency loss function, and the third efficiency loss function, and is the respective weight for the quality loss function, cost loss function, and efficiency loss function, and is the value obtained by splitting the correct answer score for the domain, and is the completeness score value for the domain, and is a weight value for calculating the above quality loss function, and is the calculated value for the token relative to the price, and is the sum of other processing costs excluding the price, and is a weight value for calculating the above cost-loss function, and is a ratio value to the actual processing time, and is a ratio value to the number of tokens used, and is a weight value for calculating the above efficiency loss function.
  16. In Article 15, The above prediction model is, A plurality of redundant data, a plurality of first text result data, a plurality of second text result data, a plurality of first image result data, a plurality of second image result data, a plurality of first redundant result data, and a plurality of second redundant result data, and additionally trained based on a plurality of total loss functions, a plurality of quality loss functions, a plurality of cost loss functions, and a plurality of efficiency loss functions, method.
  17. In Article 15, The above method is, If the above customized model is below a preset first model threshold, it is determined that the above customized model is a user-customized model, and If the customized model is greater than or equal to the first threshold value or less than or equal to the second threshold value which is set to be greater than the first threshold value, the customized model is configured to determine that some error has occurred. method.
  18. In Article 17, The above method is, If the customized model is greater than or equal to the first threshold and the second threshold, it is determined that the error of the customized model is very serious, and the prediction model is retrained only if the customized model is greater than or equal to the first threshold and the second threshold, and Through the communication interface above, the customized model is configured to be transmitted to the user's external electronic device, limited to the customized model that does not exceed the second threshold value. method.

Description

Server for recommending a user-customized model based on a predictive model and method of operation thereon Various embodiments of the present invention relate to a server that recommends a user-customized model based on a prediction model and a method for operating the same. Recently, there has been diverse interest in artificial intelligence technology, and individuals are learning about and utilizing AI in various ways. However, conventional model recommendation methods had limitations such as difficulty in making appropriate recommendations due to a lack of data on new users or new items, recommendations based on previous interactions failing to reflect changes in the user's current interests, potential for within-group bias in similar user-based recommendations, biased information exposure when recommendations are based solely on user preferences, failure to reflect current contexts such as weather, location, and emotional state, unclear methods for collecting and utilizing user data, and the fact that most existing models are text/click-based and do not properly reflect voice, haptic feedback, images, or biosignals. Therefore, there is a need for a system and method that can easily provide customized models to users while overcoming these problems. FIG. 1 illustrates a block diagram of an electronic device and network according to various embodiments of the present invention. FIG. 2 is an illustrative diagram for explaining how a server operates according to various embodiments of the present invention. FIG. 3 is an example diagram for a server to classify specific text data according to various embodiments of the present invention. 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). 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). An electronic device according to various embodiments of the present document may include, for example, at least one of a smartphone, a tablet PC, a desktop PC, a laptop PC, a netbook computer, a workstation, and a server. Referring to FIG. 1, a server (108) within a network environment (100) in various embodiments is described. The server (108) may include a bus (110), a processor (120), memory (130), an input/output interface (140), a display (150), and a communication interface (160). In some embodiments, the server (108) may omit at least one of the components or additionally include other components. The bus (110) may include a circuit that connects the components (110-160) 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 server (108). The memory (130) may include volatile and/or non-volatile memory. The memory (130) may store commands or data related to at least one other component of the server (108), for example. According to one embodiment, the memory (130) may store software and/or programs (140). The inpu