Journal of Computer Applications ›› 2023, Vol. 43 ›› Issue (3): 674-684.DOI: 10.11772/j.issn.1001-9081.2022020198

• Artificial intelligence • Previous Articles    

Survey of label noise learning algorithms based on deep learning

Boyi FU1,2, Yuncong PENG1,2, Xin LAN1,2, Xiaolin QIN1,2()   

  1. 1.Chengdu Institute of Computer Application,Chinese Academy of Sciences,Chengdu Sichuan 610041,China
    2.School of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 100049,China
  • Received:2022-02-22 Revised:2022-05-18 Accepted:2022-05-26 Online:2022-08-16 Published:2023-03-10
  • Contact: Xiaolin QIN
  • About author:FU Boyi, born in 1998, M. S. candidate. Her research interests include label noise, image semantic understanding, object detection.
    PENG Yuncong, born in 1998, M. S. candidate. His research interests include statistical machine learning, image semantic understanding, few-shot learning.
    LAN Xin, born in 1998, M. S. candidate. Her research interests include deep learning, image semantic understanding, object detection.
    QIN Xiaolin, born in 1980, Ph. D., research fellow. His research interests include automated reasoning, artificial intelligence.
  • Supported by:
    National Academy of Science Alliance Collaborative Program (Chengdu Branch of Chinese Academy of Sciences — Chongqing Academy of Science and Technology), Key Reginal Program of Science and Technology Service Network Initiative (Type A)(KFJ-STS-QYZD-2021-21-001);Sichuan Science and Technology Program(2019ZDZX0006)


伏博毅1,2, 彭云聪1,2, 蓝鑫1,2, 秦小林1,2()   

  1. 1.中国科学院 成都计算机应用研究所,成都 610041
    2.中国科学院大学 计算机科学与技术学院,北京 100049
  • 通讯作者: 秦小林
  • 作者简介:伏博毅(1998—),女,湖南岳阳人,硕士研究生,CCF会员,主要研究方向:标签噪声、图像语义理解、目标检测
  • 基金资助:


In the field of deep learning, a large number of correctly labeled samples are essential for model training. However, in practical applications, labeling data requires high labeling cost. At the same time, the quality of labeled samples is affected by subjective factors or tool and technology of manual labeling, which inevitably introduces label noise in the annotation process. Therefore, existing training data available for practical applications is subject to a certain amount of label noise. How to effectively train training data with label noise has become a research hotspot. Aiming at label noise learning algorithms based on deep learning, firstly, the source, classification and impact of label noise learning strategies were elaborated; secondly, four label noise learning strategies based on data, loss function, model and training method were analyzed according to different elements of machine learning; then, a basic framework for learning label noise in various application scenarios was provided; finally, some optimization ideas were given, and challenges and future development directions of label noise learning algorithms were proposed.

Key words: label noise, semi-supervised learning, supervised learning, deep learning, loss function



关键词: 标签噪声, 半监督学习, 监督学习, 深度学习, 损失函数

CLC Number: