[go: up one dir, main page]

WO2018131749A1 - Module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond - Google Patents

Module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond Download PDF

Info

Publication number
WO2018131749A1
WO2018131749A1 PCT/KR2017/002341 KR2017002341W WO2018131749A1 WO 2018131749 A1 WO2018131749 A1 WO 2018131749A1 KR 2017002341 W KR2017002341 W KR 2017002341W WO 2018131749 A1 WO2018131749 A1 WO 2018131749A1
Authority
WO
WIPO (PCT)
Prior art keywords
self
mission
dna
model
engine module
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/KR2017/002341
Other languages
English (en)
Korean (ko)
Inventor
윤희병
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dna System
Original Assignee
Dna System
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Dna System filed Critical Dna System
Publication of WO2018131749A1 publication Critical patent/WO2018131749A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models

Definitions

  • the present invention relates to a learning engine module, and more particularly, to a deep learning based self-adaptive learning engine module.
  • AI technology is rapidly developing through deep learning based on artificial neural networks.
  • existing artificial neural network technology or self-adaptation technology has a limitation in that it is difficult to process unstructured data because it cannot effectively implement the human brain mechanism, and it is difficult to apply to various situations because it is developed in a specific field. There is a limit that is not efficient.
  • Korean Patent Publication No. 10-2012-0057319 Invention: Intelligent sensor middleware structure that can be applied to various environments and self-adaptable for smart environment configuration, published date: June 2012 Publication No. 10-1999-0044063 (name of the invention: a method for providing a self-adaptive management service using an information communication network, publication date: May 07, 2001) and the like have been disclosed.
  • the present invention is proposed to solve the above problems of the proposed method, by combining self-adaptation technology and deep learning-based learning technology, by self-organizing the DNA mission and self-organizing the artificial neural network DNA model, Its purpose is to provide a deep learning-based self-adaptive learning engine module that can effectively understand the situation, build a model by itself, and effectively implement the human brain mechanism to solve the situation.
  • the present invention is implemented in a modular form, easy to apply to a variety of systems, self-adaptive learning using the structured data and unstructured data, utilizing the learning results to understand the situation and scheduling, decision and prediction,
  • the purpose of the present invention is to provide a deep learning-based self-adaptive learning engine module that can be applied to various recommendations and situational actions.
  • Deep learning-based self-adaptive learning engine module according to a feature of the present invention for achieving the above object
  • a mission organization unit for processing the data to self-organize the DNA mission
  • Including the model learning unit for self-learning the self-constructed DNA model is characterized by its configuration.
  • the mission organization Preferably, the mission organization, the mission organization, and
  • the mission module may be a function of elements extracted from the data and positions of organizations.
  • It may include a Special DNA Mission consisting of a combination of chains.
  • the mission organization Preferably, the mission organization, the mission organization, and
  • the tissue At least one of the blocks and chains of may be modified to self-organize DNA missions that change over time.
  • the model component Preferably, the model component,
  • the functional submodel may be a function of elements extracted from the data and sequences of thoughts.
  • the deep learning-based self-adaptive learning engine module proposed by the present invention, by combining self-adaptive technology and deep learning-based learning technology, self-organizing the DNA mission and self-organizing the artificial neural network DNA model to understand the situation You can effectively implement the human brain mechanisms to identify missions, model and solve situations.
  • the present invention is implemented in a modular form, easy to apply to a variety of systems, self-adaptive learning using the structured data and unstructured data, utilizing the learning results to understand the situation and scheduling, decision and prediction, It can be applied to recommendations and situational measures.
  • FIG. 1 is a diagram showing the configuration of a deep learning-based self-adaptive learning engine module according to an embodiment of the present invention.
  • FIG. 2 is a diagram illustrating a method of self-organizing a DNA mission in a mission organization unit of a deep learning-based self-adaptive learning engine module according to an embodiment of the present invention.
  • FIG. 3 is a view illustrating an example of a method for configuring a DNA mission in a mission organization unit of a deep learning based self-adaptive learning engine module according to an embodiment of the present invention.
  • FIG. 4 is a diagram illustrating a method of self-configuring a DNA model in a model constructing unit of a deep learning based self-adaptive learning engine module according to an embodiment of the present invention.
  • the deep learning based self-adaptive learning engine module 100 may include a mission organization unit 110, a model constructing unit 120, and a model learning unit 130. It can be configured to include.
  • the present invention combines self-adaptive technology with deep learning-based learning technology to self-organize DNA missions and self-organize artificial neural network DNA models to understand the situation by themselves, identify missions and create models to solve the situation.
  • the human brain mechanism can be effectively implemented.
  • the present invention is implemented in a modular form is easy to apply to a variety of systems.
  • the mission organization unit 110 may process data to self-organize the DNA mission.
  • the data may include not only structured data but also unstructured data such as natural language input by a person, and the mission organization unit 110 processes the data to construct a DNA mission using extracted elements. Can organize themselves.
  • a separate preprocessing module (not shown) may be included for processing data to extract elements.
  • the mission organization unit 110 may self-organize the DNA mission composed of the sum of the mission modules, and the mission module may include elements of the elements extracted from the data and positions of the positions of the organization members. It can be a function. At this time, the position of the organization member may be predetermined.
  • the DNA mission may be a combination of Blocks of Organization and Chains. That is, the mission organization unit 110 may use a block chain combination technique to organize a DNA mission by combining a block and a chain of tissue.
  • the mission organization unit 110 of the deep learning-based self-adaptive learning engine module 100 includes a block of organization (Block i, Block j, Block k, etc.) as shown in FIG. 2.
  • DNA missions can be constructed by combining chains (Chain l, Chain m, Chain n, etc.) through blockchain combination technology.
  • the constructed DNA mission may be expressed as the sum of the mission module, which is a function of the position of the element and the tissue member.
  • the DNA mission may include a special DNA mission composed of a combination of chains. That is, according to the embodiment, the DNA mission may be composed of a combination of chains without block of tissue.
  • FIG. 3 is a diagram illustrating a method of constructing a DNA mission over time in the mission organization unit 110 of the deep learning based self-adaptive learning engine module 100 according to an embodiment of the present invention.
  • the mission organization unit 110 of the deep learning-based self-adaptive learning engine module 100 is extracted from elements and data within a mission and mission of a predetermined organization. By comparing and evaluating elements, you can organize your own DNA missions that change over time.
  • the mission organization unit 110 may change the DNA mission of the organization at a specific time point (1), if the mission of the organization member is changed or the member itself is changed at a specific time point (2), and in the passage of time. Therefore, when the element exceeds a predetermined limit (3), at least one of the block and chain of the tissue can be changed to self-organize the DNA mission that changes over time.
  • the mission organization unit 110 changes and combines both the block and the chain, Self-organizing DNA missions can change over time.
  • the mission organization unit 110 keeps the block of the organization as it is. By altering and recombining the chain, the DNA mission can be self-organized.
  • an element exceeds a predetermined threshold value (3) as time passes, the mission organization unit 110 automatically leaves the block of tissue intact. Chains can be altered and combined to self-organize DNA missions that change appropriately over time.
  • the model constructing unit 120 may self-configure the deep learning-based artificial neural network DNA model using a self-organized DNA mission. That is, the model constructer 120 may receive a DNA mission from the mission organizer 110 as an input and construct a DNA model that can be trained on a deep learning basis. As such, the model constructing unit 120 of the deep learning-based self-adaptive learning engine module 100 according to an embodiment of the present invention may generate a DNA model using a DNA mission that is self-organized using elements extracted from data. Because of the configuration, the model can be flexibly changed in accordance with the input data.
  • model constructor 120 self-constructs a DNA model consisting of a sum of functional submodels, and the submodels are functions of elements extracted from data and sequences of thoughts. Can be.
  • the DNA model may be a combination of blocks of function and chains. That is, the model constructing unit 120 may use a block chain combination technique to organize a DNA model by combining a functional block and a chain.
  • FIG. 4 is a diagram illustrating a method of self-configuring a DNA model in the model constructing unit 120 of the deep learning-based self-adaptive learning engine module 100 according to an embodiment of the present invention.
  • Combination technologies can be combined to form functional submodels (Functional Submodel i, Functional Submodel j, Functional Submodel k, Functional Submodel m, Functional Submodel n, etc.) and to construct DNA models from the sum of the functional submodels.
  • the model learner 130 may self-learn a self-constructed DNA model. That is, the model learning unit 130 is configured to learn the DNA model configured in the model constructing unit 120, and can learn through an artificial neural network technology, and utilizes the learning results of the model learning unit 130. It can be applied to understanding and scheduling, decision making and prediction, recommendation and situational action.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Genetics & Genomics (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)

Abstract

La présente invention concerne un module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond selon lequel le mécanisme du cerveau humain pour identifier une mission de lui-même en comprenant une situation en résolvant la situation en créant un modèle peut être efficacement mis en œuvre en auto-organisant une mission d'ADN et auto-configurant un modèle d'ADN de réseau neuronal artificiel en combinant une technologie auto-adaptative et une technologie d'apprentissage basée sur l'apprentissage profond. En outre, la présente invention est mise en œuvre sous forme de module et peut donc être facilement appliquée sur divers systèmes et, en réalisant un apprentissage auto-adaptatif à l'aide de données structurées et de données non structurées, telles que la compréhension d'une situation, la planification, la prise de décision, la prédiction, la recommandation et la prise d'une mesure selon une situation, en utilisant le résultat d'apprentissage.
PCT/KR2017/002341 2017-01-16 2017-03-03 Module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond Ceased WO2018131749A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR20170007354 2017-01-16
KR10-2017-0007354 2017-01-16

Publications (1)

Publication Number Publication Date
WO2018131749A1 true WO2018131749A1 (fr) 2018-07-19

Family

ID=60296273

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2017/002341 Ceased WO2018131749A1 (fr) 2017-01-16 2017-03-03 Module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond

Country Status (2)

Country Link
KR (1) KR101787611B1 (fr)
WO (1) WO2018131749A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110245843A (zh) * 2019-05-24 2019-09-17 深圳市元征科技股份有限公司 一种基于区块链的信息管理的方法及相关装置

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109698822A (zh) * 2018-11-28 2019-04-30 众安信息技术服务有限公司 基于公有区块链和加密神经网络的联合学习方法及系统
KR102559590B1 (ko) 2020-05-29 2023-07-26 한국전자통신연구원 신경망의 불확실성에 기반한 지식 증강 방법 및 장치
KR20220081782A (ko) 2020-12-09 2022-06-16 삼성전자주식회사 뉴럴 네트워크를 이용하는 데이터 처리 방법, 데이터 처리 장치 및 이를 포함한 전자 장치

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003317073A (ja) * 2002-04-24 2003-11-07 Fuji Xerox Co Ltd ニューラルネットワーク処理装置
JP2005182449A (ja) * 2003-12-19 2005-07-07 Takumi Ichimura ニューラルネットワークによるデータベース解析装置
KR20060076839A (ko) * 2004-12-29 2006-07-05 학교법인 대양학원 상황인식 서비스를 제공하는 장치 및 방법
KR20070043126A (ko) * 2005-10-20 2007-04-25 (주)아라게이트 게임용 인공지능 엔진시스템
KR101607209B1 (ko) * 2014-10-16 2016-03-30 아주대학교산학협력단 온톨로지를 이용한 자가적응시스템 및 자가적응방법

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008041085A (ja) * 2006-07-14 2008-02-21 Pacific Technos Corp 環境に対して自律的に適応する自律適応型システム

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003317073A (ja) * 2002-04-24 2003-11-07 Fuji Xerox Co Ltd ニューラルネットワーク処理装置
JP2005182449A (ja) * 2003-12-19 2005-07-07 Takumi Ichimura ニューラルネットワークによるデータベース解析装置
KR20060076839A (ko) * 2004-12-29 2006-07-05 학교법인 대양학원 상황인식 서비스를 제공하는 장치 및 방법
KR20070043126A (ko) * 2005-10-20 2007-04-25 (주)아라게이트 게임용 인공지능 엔진시스템
KR101607209B1 (ko) * 2014-10-16 2016-03-30 아주대학교산학협력단 온톨로지를 이용한 자가적응시스템 및 자가적응방법

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110245843A (zh) * 2019-05-24 2019-09-17 深圳市元征科技股份有限公司 一种基于区块链的信息管理的方法及相关装置
CN110245843B (zh) * 2019-05-24 2023-08-18 深圳市元征科技股份有限公司 一种基于区块链的信息管理的方法及相关装置

Also Published As

Publication number Publication date
KR101787611B1 (ko) 2017-10-18

Similar Documents

Publication Publication Date Title
WO2018131749A1 (fr) Module de moteur d'apprentissage auto-adaptatif basé sur l'apprentissage profond
WO2018135696A1 (fr) Plate-forme d'intelligence artificielle utilisant une technologie d'apprentissage auto-adaptative basée sur apprentissage profond
WO2022163996A1 (fr) Dispositif pour prédire une interaction médicament-cible à l'aide d'un modèle de réseau neuronal profond à base d'auto-attention, et son procédé
WO2021096009A1 (fr) Procédé et dispositif permettant d'enrichir la connaissance sur la base d'un réseau de relations
WO2021132797A1 (fr) Procédé de classification d'émotions de parole dans une conversation à l'aide d'une incorporation d'émotions mot par mot, basée sur un apprentissage semi-supervisé, et d'un modèle de mémoire à court et long terme
WO2019066104A1 (fr) Procédé et système de commande de traitement faisant appel à un apprentissage de réseau neuronal basé sur des données d'historique
WO2021095987A1 (fr) Procédé et appareil de complémentation de connaissances basée sur une entité de type multiple
WO2018212394A1 (fr) Procédé, dispositif et programme informatique pour l'exploitation d'un environnement d'apprentissage automatique
WO2017164478A1 (fr) Procédé et appareil de reconnaissance de micro-expressions au moyen d'une analyse d'apprentissage profond d'une dynamique micro-faciale
WO2016159497A1 (fr) Procédé, système et support d'enregistrement lisible par ordinateur non transitoire pour la présentation d'informations d'apprentissage
WO2019059493A1 (fr) Système de relation avec l'utilisateur utilisant un agent conversationnel
WO2022260392A1 (fr) Procédé et système pour générer un modèle de réseau neuronal artificiel de traitement d'image fonctionnant dans un terminal
WO2021107422A1 (fr) Procédé de surveillance de charge non intrusive utilisant des données de consommation d'énergie
WO2024019474A1 (fr) Onduleur bidirectionnel à fonction d'onduleur solaire
WO2018143486A1 (fr) Procédé de fourniture de contenu utilisant un système de modularisation pour analyse d'apprentissage profond
WO2023224205A1 (fr) Procédé de génération de modèle commun par synthèse de résultat d'apprentissage de modèle de réseau neuronal artificiel
WO2023128093A1 (fr) Appareil et procédé d'apprentissage par renforcement basés sur un environnement d'apprentissage utilisateur dans la conception de semi-conducteur
WO2022139325A1 (fr) Système informatique pour apprentissage adaptatif multi-domaine basé sur un réseau neuronal unique sans sur-apprentissage, et procédé associé
WO2023106466A1 (fr) Dispositif et procédé d'apprentissage en nuage d'intelligence artificielle basé sur un type de nuage d'apprentissage
WO2018151366A1 (fr) Module d'auto-composition de modèle de réseau neuronal artificiel utilisant une combinaison de chaînes de neuro-blocs
WO2018212398A1 (fr) Procédé d'analyse d'exploration d'esprit utilisant un lien entre des données de visualisation
WO2023022406A1 (fr) Procédé d'évaluation de capacité d'apprentissage, dispositif d'évaluation de capacité d'apprentissage et système d'évaluation de capacité d'apprentissage
WO2023101117A1 (fr) Procédé de gestion d'enseignement en distanciel à l'aide d'une reconnaissance de personne basée sur l'apprentissage profond
WO2018147495A1 (fr) Module destine a l'auto-organisation d'une mission à l'aide de combinaison neuro-bloc-chaîne
WO2025048065A1 (fr) Procédé et appareil de collecte de données par reconnaissance d'écran ihm et analyse d'image à l'aide d'une caméra

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17891825

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 211019)

122 Ep: pct application non-entry in european phase

Ref document number: 17891825

Country of ref document: EP

Kind code of ref document: A1