WO2024090600A1 - Procédé d'entrainement de modèle d'apprentissage profond et appareil de calcul d'apprentissage profond appliqué à celui-ci - Google Patents

Procédé d'entrainement de modèle d'apprentissage profond et appareil de calcul d'apprentissage profond appliqué à celui-ci Download PDF

Info

Publication number
WO2024090600A1
WO2024090600A1 PCT/KR2022/016397 KR2022016397W WO2024090600A1 WO 2024090600 A1 WO2024090600 A1 WO 2024090600A1 KR 2022016397 W KR2022016397 W KR 2022016397W WO 2024090600 A1 WO2024090600 A1 WO 2024090600A1
Authority
WO
WIPO (PCT)
Prior art keywords
deep learning
weights
learning model
pruning
loading
Prior art date
Application number
PCT/KR2022/016397
Other languages
English (en)
Korean (ko)
Inventor
이상설
장성준
김경호
Original Assignee
한국전자기술연구원
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 한국전자기술연구원 filed Critical 한국전자기술연구원
Publication of WO2024090600A1 publication Critical patent/WO2024090600A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
    • G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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

Definitions

  • the present invention relates to image-based deep learning processing and system SoC (System on chip) technology, and more specifically, to a method of learning a deep learning model at high speed with high accuracy in a lightweight deep learning computing device.
  • SoC System on chip
  • the present invention was created to solve the above problems, and the purpose of the present invention is to quickly learn a deep learning model using additional datasets in a deep learning computing device with limited resources, while maintaining a high level of prediction accuracy. To provide a deep learning model learning method that can be maintained and a deep learning computing device to which it is applied.
  • a deep learning model learning method to achieve the above object includes a first learning step of training a deep learning model; A first pruning step of pruning some weights in the learned deep learning model; It includes a first loading step of loading specific weights into the pruned weights.
  • the first loading step may load the weights of a previously learned deep learning model.
  • the deep learning model to which the weights of the previously learned deep learning model are transferred may be fine-tuned to the first data set.
  • a deep learning model learning method includes a second learning step of fine tuning the deep learning model on which the first loading step has been performed with a second data set; A second pruning step of pruning some weights in the fine-tuned deep learning model; It may further include a second loading step of loading specific weights into the pruned weights.
  • the second loading step may load the weights of a previously learned deep learning model. Some weights pruned in the second pruning step may be some of the weights pruned in the first pruning step.
  • the first pruning step and the second pruning step may prune weights on a channel basis.
  • the first pruning step and the second pruning step may prune weights of different channels for each layer.
  • Deep learning models can be mounted on lightweight, low-power deep learning computing devices.
  • a deep learning computing device trains a deep learning model. An operator that prunes some weights from the learned deep learning model and loads specific weights into the pruned weights; and a memory that provides storage space required for the calculator.
  • a deep learning model learning method includes a first pruning step of pruning some weights in the deep learning model; A first loading step of loading specific weights into the pruned weights; a second pruning step of pruning some weights in the deep learning model in which the first loading step was performed; It includes a second loading step of loading specific weights into the pruned weights.
  • a deep learning computing device prunes some weights in a deep learning model, loads specific weights on the pruned weights, and prunes some weights in the deep learning model loaded with specific weights.
  • 1 is a diagram conceptually showing a deep learning model learning method in a deep learning computing device
  • Figure 2 shows test results for the transfer learned deep learning model
  • 3 to 5 are diagrams provided to explain a deep learning model learning method according to an embodiment of the present invention.
  • Figure 6 is a diagram showing the configuration of a deep learning computing device according to another embodiment of the present invention.
  • Figure 1 is a diagram conceptually showing a deep learning model learning method in a deep learning computing device (deep learning accelerator). As shown in the upper part of FIG. 1, a deep learning computing device that cannot learn from many learning datasets provides additional data for the deep learning model transfer learned by the server as shown in the lower part of FIG. 1. It is carried out by learning three.
  • Figure 2 shows test results for the transfer learned deep learning model. As shown, when a transfer-learned deep learning model is additionally trained, learning performance quickly increases compared to a deep learning model without transfer learning.
  • FC layer Fely Connected Layer
  • An embodiment of the present invention presents a deep learning model learning method that can quickly train a deep learning model using an additional dataset in a deep learning computing device with limited resources while maintaining high prediction accuracy.
  • 3 to 5 are diagrams provided to explain a deep learning model learning method according to an embodiment of the present invention.
  • the deep learning model learning method according to an embodiment of the present invention is suitable for learning a deep learning model mounted on a lightweight deep learning accelerator, but is not necessarily limited to this and can also be applied in other environments/methods.
  • weights are transferred to the deep learning model as shown in Figure 3. This is a process of securing the weights of the deep learning model acquired through pre-training using a large amount of learning data sets at the server side and loading them into the deep learning model to be learned.
  • the weights shown on the left are the weights of the first layer, and the weights shown on the right are the weights of the second layer.
  • the deep learning model trained in the embodiment of the present invention consists of two layers, but this is only an example for convenience of explanation. There is no limit to the number of layers of a deep learning model to which embodiments of the present invention can be applied.
  • the deep learning model has a structure in which images are input through multi-channels, and feature maps of the images are also generated through multi-channels, and are divided by weights for each channel.
  • the deep learning accelerator uses dataset #1 to fine-tune the deep learning model to which the weights have been transferred, and to select weights subject to pruning.
  • weights subject to pruning are those displayed in white.
  • weight pruning is performed on a channel basis. That is, the weights for some channels are pruned and the weights for the remaining channels are left. Meanwhile, weight pruning can prune the weights of different channels for each layer. As shown, the weight pruning target channels in the first layer shown on the left and the weight pruning target channels in the second layer shown on the right are different from each other.
  • the weights of the previously learned deep learning model are loaded for the pruned weights.
  • Previously 0 was loaded into pruned waddles or randomly generated weights were loaded.
  • the prediction accuracy of the deep learning model was improved by loading the weights of the previously learned deep learning model into the pruned weights.
  • the deep learning accelerator uses dataset #2 to fine-tune the deep learning model that has been learned through the process shown in FIG. 4, and select weights subject to pruning. .
  • weights that were not subject to pruning in FIG. 4 can be excluded from the pruning subject. That is, among the weights that were the pruning target in FIG. 4, some weights are selected as the pruning target. Weights that were not subject to pruning in Figure 4 are not selected for pruning even in learning using dataset #2.
  • weights selected from the fine-tuned deep learning model are pruned.
  • the weights subject to pruning are those displayed in white.
  • weight pruning is performed on a channel basis, and the pruning target channel for each layer may be different. .
  • the weights of the previously learned deep learning model are loaded onto the pruned weights.
  • Figure 6 is a diagram showing the configuration of a deep learning computing device according to another embodiment of the present invention.
  • the deep learning computing device includes a communication interface 110, a deep learning calculator 120, and a memory 130.
  • the communication interface 110 communicates with an external host system and receives data sets, parameters (weight, bias) of previously learned deep learning models, etc.
  • the deep learning calculator 120 trains the mounted deep learning model using the method shown in FIGS. 3 to 5 described above.
  • the memory 130 provides storage space necessary for the deep learning calculator 120 to perform calculations.
  • the deep learning processing unit does not perform calculations on the pruned weights, allowing high-speed learning with low power while maintaining prediction accuracy at a high level.
  • a computer-readable recording medium can be any data storage device that can be read by a computer and store data.
  • computer-readable recording media can be ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, etc.
  • computer-readable codes or programs stored on a computer-readable recording medium may be transmitted through a network connected between computers.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Neurology (AREA)
  • Image Processing (AREA)
  • Feedback Control In General (AREA)
  • Manipulator (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

L'invention concerne un procédé d'entraînement de modèle d'apprentissage profond et un appareil de calcul d'apprentissage profond appliqué à celui-ci. Le procédé d'entraînement de modèle d'apprentissage profond selon un mode de réalisation de la présente invention comprend l'entraînement d'un modèle d'apprentissage profond, l'élagage de certains poids dans le modèle d'apprentissage profond entraîné, et le chargement de poids spécifiques sur les poids élagués. En conséquence, l'appareil de calcul d'apprentissage profond dans lequel des ressources sont limitées peut effectuer rapidement un entraînement tout en améliorant rapidement la précision de prédiction en appliquant des poids pré-appris à des poids élagués pendant un entraînement de modèle d'apprentissage profond par un ensemble de données supplémentaire.
PCT/KR2022/016397 2022-10-26 2022-10-26 Procédé d'entrainement de modèle d'apprentissage profond et appareil de calcul d'apprentissage profond appliqué à celui-ci WO2024090600A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020220138798A KR20240058252A (ko) 2022-10-26 2022-10-26 딥러닝 모델 학습 방법 및 이를 적용한 딥러닝 연산장치
KR10-2022-0138798 2022-10-26

Publications (1)

Publication Number Publication Date
WO2024090600A1 true WO2024090600A1 (fr) 2024-05-02

Family

ID=90831078

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2022/016397 WO2024090600A1 (fr) 2022-10-26 2022-10-26 Procédé d'entrainement de modèle d'apprentissage profond et appareil de calcul d'apprentissage profond appliqué à celui-ci

Country Status (2)

Country Link
KR (1) KR20240058252A (fr)
WO (1) WO2024090600A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20180013674A (ko) * 2016-07-28 2018-02-07 삼성전자주식회사 뉴럴 네트워크의 경량화 방법, 이를 이용한 인식 방법, 및 그 장치
KR20210015990A (ko) * 2019-05-18 2021-02-10 주식회사 디퍼아이 학습된 파라미터의 형태변환을 이용한 컨벌루션 신경망 파라미터 최적화 방법, 컨벌루션 신경망 연산방법 및 그 장치
KR20210108413A (ko) * 2018-12-18 2021-09-02 모비디어스 리미티드 뉴럴 네트워크 압축
KR20220085280A (ko) * 2020-12-15 2022-06-22 경희대학교 산학협력단 초해상화를 수행하는 인공 신경망의 가중치를 처리하는 방법 및 장치
KR20220116270A (ko) * 2020-02-07 2022-08-22 주식회사 히타치하이테크 학습 처리 장치 및 방법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20180013674A (ko) * 2016-07-28 2018-02-07 삼성전자주식회사 뉴럴 네트워크의 경량화 방법, 이를 이용한 인식 방법, 및 그 장치
KR20210108413A (ko) * 2018-12-18 2021-09-02 모비디어스 리미티드 뉴럴 네트워크 압축
KR20210015990A (ko) * 2019-05-18 2021-02-10 주식회사 디퍼아이 학습된 파라미터의 형태변환을 이용한 컨벌루션 신경망 파라미터 최적화 방법, 컨벌루션 신경망 연산방법 및 그 장치
KR20220116270A (ko) * 2020-02-07 2022-08-22 주식회사 히타치하이테크 학습 처리 장치 및 방법
KR20220085280A (ko) * 2020-12-15 2022-06-22 경희대학교 산학협력단 초해상화를 수행하는 인공 신경망의 가중치를 처리하는 방법 및 장치

Also Published As

Publication number Publication date
KR20240058252A (ko) 2024-05-03

Similar Documents

Publication Publication Date Title
CN106960219A (zh) 图片识别方法及装置、计算机设备及计算机可读介质
CN106815311B (zh) 一种问题匹配方法和装置
WO2021125619A1 (fr) Procédé d'inspection de marquage sur zone de délimitation à l'aide d'un modèle d'apprentissage profond et appareil faisant appel audit procédé
CN1943179A (zh) 用于动态地可扩展的虚拟交换机的装置和方法
CN107391549A (zh) 基于人工智能的新闻召回方法、装置、设备及存储介质
WO2021118041A1 (fr) Procédé pour distribuer un travail d'étiquetage en fonction de sa difficulté, et appareil l'utilisant
CN108229535A (zh) 涉黄图像审核方法、装置、计算机设备及存储介质
WO2024090600A1 (fr) Procédé d'entrainement de modèle d'apprentissage profond et appareil de calcul d'apprentissage profond appliqué à celui-ci
WO2022146080A1 (fr) Algorithme et procédé de modification dynamique de la précision de quantification d'un réseau d'apprentissage profond
CN109815992A (zh) 一种支持向量机并行加速训练方法及系统
WO2022107925A1 (fr) Dispositif de traitement de détection d'objet à apprentissage profond
WO2023033194A1 (fr) Procédé et système de distillation de connaissances spécialisés pour l'éclaircissement de réseau neuronal profond à base d'élagage
WO2023085458A1 (fr) Procédé et dispositif de commande d'une mémoire de formation d'apprentissage profond allégée
WO2024135867A1 (fr) Procédé d'apprentissage à transfert efficace pour réseau d'apprentissage profond de petite échelle
WO2022107927A1 (fr) Appareil d'apprentissage profond permettant un post-traitement rapide
WO2022107951A1 (fr) Procédé de formation d'un réseau d'apprentissage profond ultra-léger
WO2024135860A1 (fr) Procédé d'élagage de données pour dispositif matériel léger d'apprentissage profond
WO2022102912A1 (fr) Procédé de sélection dynamique d'architecture neuromorphique pour la modélisation sur la base d'un paramètre de modèle snn, et support d'enregistrement et dispositif pour son exécution
WO2023095934A1 (fr) Procédé et système d'allégement d'un réseau neuronal à tête d'un détecteur d'objet
WO2023113450A1 (fr) Procédé d'application de dissipateur de support pour analyse de dissipation de chaleur d'impression 3d
WO2024135862A1 (fr) Dispositif de traitement et de manipulation de données prenant en charge un traitement de données non structurées
WO2024135861A1 (fr) Procédé d'entraînement de réseau d'apprentissage profond appliquant un type de représentation de données variable, et dispositif mobile l'appliquant
WO2022107929A1 (fr) Accélérateur d'apprentissage profond comprenant un compresseur/décompresseur de données variables
WO2023080291A1 (fr) Dispositif de regroupement pour accélérateur d'apprentissage profond
CN112819022B (zh) 基于神经网络的图像识别装置和图像识别方法

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: 22963561

Country of ref document: EP

Kind code of ref document: A1