Contrasting Contrastive Self-Supervised Representation Learning Pipelines

被引:10
|
作者
Kotar, Klemen [1 ]
Ilharco, Gabriel [2 ]
Schmidt, Ludwig [2 ]
Ehsani, Kiana [1 ]
Mottaghi, Roozbeh [1 ,2 ]
机构
[1] PRIOR Allen Inst AI, Seattle, WA 98103 USA
[2] Univ Washington, Seattle, WA 98195 USA
关键词
D O I
10.1109/ICCV48922.2021.00980
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In the past few years, we have witnessed remarkable breakthroughs in self-supervised representation learning. Despite the success and adoption of representations learned through this paradigm, much is yet to be understood about how different training methods and datasets influence performance on downstream tasks. In this paper, we analyze contrastive approaches as one of the most successful and popular variants of self-supervised representation learning. We perform this analysis from the perspective of the training algorithms, pre-training datasets and end tasks. We examine over 700 training experiments including 30 encoders, 4 pre-training datasets and 20 diverse downstream tasks. Our experiments address various questions regarding the performance of self-supervised models compared to their supervised counterparts, current benchmarks used for evaluation, and the effect of the pre-training data on end task performance. Our Visual Representation Benchmark (ViRB) is available at: https://github.com/allenai/virb.
引用
收藏
页码:9929 / 9939
页数:11
相关论文
共 50 条
  • [31] Contrastive self-supervised representation learning framework for metal surface defect detection
    Mahe Zabin
    Anika Nahian Binte Kabir
    Muhammad Khubayeeb Kabir
    Ho-Jin Choi
    Jia Uddin
    Journal of Big Data, 10
  • [32] Generative Variational-Contrastive Learning for Self-Supervised Point Cloud Representation
    Wang, Bohua
    Tian, Zhiqiang
    Ye, Aixue
    Wen, Feng
    Du, Shaoyi
    Gao, Yue
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2024, 46 (09) : 6154 - 6166
  • [33] TimeCLR: A self-supervised contrastive learning framework for univariate time series representation
    Yang, Xinyu
    Zhang, Zhenguo
    Cui, Rongyi
    KNOWLEDGE-BASED SYSTEMS, 2022, 245
  • [34] Contrastive self-supervised representation learning framework for metal surface defect detection
    Zabin, Mahe
    Kabir, Anika Nahian Binte
    Kabir, Muhammad Khubayeeb
    Choi, Ho-Jin
    Uddin, Jia
    JOURNAL OF BIG DATA, 2023, 10 (01)
  • [35] Attentive spatial-temporal contrastive learning for self-supervised video representation
    Yang, Xingming
    Xiong, Sixuan
    Wu, Kewei
    Shan, Dongfeng
    Xie, Zhao
    IMAGE AND VISION COMPUTING, 2023, 137
  • [36] CARLA: Self-supervised contrastive representation learning for time series anomaly detection
    Darban, Zahra Zamanzadeh
    Webb, Geoffrey I.
    Pan, Shirui
    Aggarwal, Charu C.
    Salehi, Mahsa
    PATTERN RECOGNITION, 2025, 157
  • [37] IPCL: ITERATIVE PSEUDO-SUPERVISED CONTRASTIVE LEARNING TO IMPROVE SELF-SUPERVISED FEATURE REPRESENTATION
    Kumar, Sonal
    Phukan, Anirudh
    Sur, Arijit
    2024 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, ICASSP 2024, 2024, : 6270 - 6274
  • [38] Self-Supervised Contrastive Molecular Representation Learning with a Chemical Synthesis Knowledge Graph
    Xie, Jiancong
    Wang, Yi
    Rao, Jiahua
    Zheng, Shuangjia
    Yang, Yuedong
    JOURNAL OF CHEMICAL INFORMATION AND MODELING, 2024, 64 (06) : 1945 - 1954
  • [39] Contrastive Spatio-Temporal Pretext Learning for Self-Supervised Video Representation
    Zhang, Yujia
    Po, Lai-Man
    Xu, Xuyuan
    Liu, Mengyang
    Wang, Yexin
    Ou, Weifeng
    Zhao, Yuzhi
    Yu, Wing-Yin
    THIRTY-SIXTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTY-FOURTH CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE / THE TWELVETH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2022, : 3380 - 3389
  • [40] Cut-in maneuver detection with self-supervised contrastive video representation learning
    Yagiz Nalcakan
    Yalin Bastanlar
    Signal, Image and Video Processing, 2023, 17 : 2915 - 2923