Journal of Computer Applications ›› 2021, Vol. 41 ›› Issue (12): 3409-3418.DOI: 10.11772/j.issn.1001-9081.2021060895

• The 18th China Conference on Machine Learning •     Next Articles

Research advances in disentangled representation learning

Keyang CHENG1,2(), Chunyun MENG1, Wenshan WANG2, Wenxi SHI2,3, Yongzhao ZHAN1   

  1. 1.School of Computer Science and Communication Engineering,Jiangsu University,Zhenjiang Jiangsu 212013,China
    2.National Engineering Laboratory of Big Data Application for Social Security Risk Perception and Prevention by Big Data (China Academy of Electronic and Information Technology),Beijing 100041,China
    3.Xinjiang Lianhai Chuangzhi Information Technology Company Limited,Urumqi Xinjiang 830011,China
  • Received:2021-05-12 Revised:2021-06-21 Accepted:2021-06-25 Online:2021-08-20 Published:2021-12-10
  • Contact: Keyang CHENG
  • About author:MENG Chunyun, born in 1994, M. S. candidate. His research interests include computer vision, pattern recognition.
    WANG Wenshan, born in 1994, M. S. Her research interests include statistical analysis, machine learning.
    SHI Wenxi, born in 1988, Ph. D. His research interests include big data analysis, smart security.
    ZHAN Yongzhao, born in 1962, Ph. D., professor. His research interest include multimedia, artificial intelligence.
  • Supported by:
    the National Natural Science Foundation of China(61972183);the Director Foundation of National Engineering Laboratory for Public Safety Risk Perception and Control by Big Data

解耦表征学习研究进展

成科扬1,2(), 孟春运1, 王文杉2, 师文喜2,3, 詹永照1   

  1. 1.江苏大学 计算机科学与通信工程学院,江苏 镇江 212013
    2.社会安全风险感知与防控大数据应用国家工程实验室(中国电子科学研究院),北京 100041
    3.新疆联海创智信息科技有限公司,乌鲁木齐 830011
  • 通讯作者: 成科扬
  • 作者简介:孟春运(1994—),男,江苏扬州人,硕士研究生,主要研究方向:计算机视觉、模式识别
    王文杉(1994—),女,湖北武汉人,硕士,主要研究方向:统计分析、机器学习
    师文喜(1988—),男,北京人,博士,主要研究方向:大数据分析、智慧安防
    詹永照(1962—),男,福建尤溪人,教授,博士生导师,博士,主要研究方向:多媒体、人工智能。
  • 基金资助:
    国家自然科学基金资助项目(61972183);社会安全风险感知与防控大数据应用国家工程实验室主任基金资助项目

Abstract:

The purpose of disentangled representation learning is to model the key factors that affect the form of data, so that the change of a key factor only causes the change of data on a certain feature, while the other features are not affected. It is conducive to face the challenge of machine learning in model interpretability, object generation and operation, zero-shot learning and other issues. Therefore, disentangled representation learning always be a research hotspot in the field of machine learning. Starting from the history and motives of disentangled representation learning, the research status and applications of disentangled representation learning were summarized, the invariance, reusability and other characteristics of disentangled representation learning were analyzed, and the research on the factors of variation via generative entangling, the research on the factors of variation with manifold interaction, and the research on the factors of variation using adversarial training were introduced, as well as the latest research trends such as a Variational Auto-Encoder (VAE) named β-VAE were introduced. At the same time, the typical applications of disentangled representation learning were shown, and the future research directions were prospected.

Key words: disentangled learning, representation learning, variational inference, interpretability, machine learning, auto-encoder, factors of variation, reusability

摘要:

解耦表征学习旨在对影响数据形态的关键因素进行建模,使得某一关键因素的变化仅仅引起数据在某项特征上的变化,而其他的特征不受影响,这有利于应对机器学习在模型可解释性、对象生成和操作以及零样本学习等问题上的挑战,因此解耦表征学习一直是机器学习领域的一个研究热点。从解耦表征学习的历史与动机入手,对解耦表征学习的研究现状以及应用进行归纳总结,分析了解耦表征所具有的不变性、复用性等特性,介绍了基于生成解耦表征变差因素的研究、基于流形相互作用解耦表征变差因素的研究、基于对抗性训练解耦表征变差因素的研究,以及一种变分自编码器β-VAE的研究等最新研究动态。同时,阐述了解耦表征学习的典型应用,并对未来的研究方向作出了展望。

关键词: 解耦学习, 表征学习, 变分推断, 可解释性, 机器学习, 自编码器, 变差因素, 复用性

CLC Number: