MarginMatch: Improving Semi-Supervised Learning with Pseudo-Margins

被引:10
|
作者
Sosea, Tiberiu [1 ]
Caragea, Cornelia [1 ]
机构
[1] Univ Illinois, Chicago, IL 60680 USA
来源
2023 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR) | 2023年
关键词
D O I
10.1109/CVPR52729.2023.01514
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We introduce MarginMatch, a new SSL approach combining consistency regularization and pseudo-labeling, with its main novelty arising from the use of unlabeled data training dynamics to measure pseudo-label quality. Instead of using only the model's confidence on an unlabeled example at an arbitrary iteration to decide if the example should be masked or not, MarginMatch also analyzes the behavior of the model on the pseudo-labeled examples as the training progresses, to ensure low quality predictions are masked out. MarginMatch brings substantial improvements on four vision benchmarks in low data regimes and on two large-scale datasets, emphasizing the importance of enforcing high-quality pseudo-labels. Notably, we obtain an improvement in error rate over the state-of-the-art of 3.25% on CIFAR-100 with only 25 labels per class and of 3.78% on STL-10 using as few as 4 labels per class. We make our code available at https://github.com/tsosea2/MarginMatch.
引用
收藏
页码:15773 / 15782
页数:10
相关论文
共 50 条
  • [21] Boosting semi-supervised network representation learning with pseudo-multitasking
    Wang, Biao
    Dai, Zhen
    Kong, Deshun
    Yu, Lanlan
    Zheng, Jin
    Li, Ping
    APPLIED INTELLIGENCE, 2022, 52 (07) : 8118 - 8133
  • [22] Semi-Supervised Adversarial Learning for Improving the Diagnosis of Pulmonary Nodules
    Fu, Yu
    Xue, Peng
    Xiao, Taohui
    Zhang, Zhili
    Zhang, Youren
    Dong, Enqing
    IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, 2023, 27 (01) : 109 - 120
  • [23] IMPROVING SMALL CONVOLUTIONAL NEURAL NETWORKS WITH SEMI-SUPERVISED LEARNING
    Badea, Mihai
    Vertan, Constantin
    Florea, Corneliu
    Florea, Laura
    Racoviţeanu, Andrei
    UPB Scientific Bulletin, Series C: Electrical Engineering and Computer Science, 2022, 84 (03): : 107 - 118
  • [24] GraphixMatch: Improving semi-supervised learning for graph classification with FixMatch
    Koh, Eunji
    Lee, Young Jae
    Kim, Seoung Bum
    NEUROCOMPUTING, 2024, 607
  • [25] IMPROVING SMALL CONVOLUTIONAL NEURAL NETWORKS WITH SEMI-SUPERVISED LEARNING
    Badea, Mihai
    Vertan, Constantin
    Florea, Corneliu
    Florea, Laura
    Racoviteanu, Andrei
    UNIVERSITY POLITEHNICA OF BUCHAREST SCIENTIFIC BULLETIN SERIES C-ELECTRICAL ENGINEERING AND COMPUTER SCIENCE, 2022, 84 (03): : 107 - 118
  • [26] FocalMatch: Mitigating Class Imbalance of Pseudo Labels in Semi-Supervised Learning
    Deng, Yongkun
    Zhang, Chenghao
    Yang, Nan
    Chen, Huaming
    APPLIED SCIENCES-BASEL, 2022, 12 (20):
  • [27] Weighted Pseudo Labeled Data and Mutual Learning for Semi-Supervised Classification
    Mo, Jianwen
    Gan, Yuwan
    Yuan, Hua
    IEEE ACCESS, 2021, 9 : 36522 - 36534
  • [28] Boosting semi-supervised network representation learning with pseudo-multitasking
    Biao Wang
    Zhen Dai
    Deshun Kong
    Lanlan Yu
    Jin Zheng
    Ping Li
    Applied Intelligence, 2022, 52 : 8118 - 8133
  • [29] Multiview Pseudo-Labeling for Semi-supervised Learning from Video
    Xiong, Bo
    Fan, Haoqi
    Grauman, Kristen
    Feichtenhofer, Christoph
    2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 7189 - 7199
  • [30] REVISITING AND IMPROVING SEMI-SUPERVISED LEARNING: A LARGE DIMENSIONAL APPROACH
    Mai, Xiaoyi
    Couillet, Romain
    2019 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2019, : 3547 - 3551