DSV: An Alignment Validation Loss for Self-supervised Outlier Model Selection

被引:0
作者
Yoo, Jaemin [1 ]
Zhao, Yue [1 ]
Zhao, Lingxiao [1 ]
Akoglu, Leman [1 ]
机构
[1] Carnegie Mellon Univ, Pittsburgh, PA 15213 USA
来源
MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES: RESEARCH TRACK, ECML PKDD 2023, PT I | 2023年 / 14169卷
基金
美国安德鲁·梅隆基金会;
关键词
Anomaly detection; Self-supervised learning; Unsupervised model selection; Data augmentation;
D O I
10.1007/978-3-031-43412-9_15
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Self-supervised learning (SSL) has proven effective in solving various problems by generating internal supervisory signals. Unsupervised anomaly detection, which faces the high cost of obtaining true labels, is an area that can greatly benefit from SSL. However, recent literature suggests that tuning the hyperparameters (HP) of data augmentation functions is crucial to the success of SSL-based anomaly detection (SSAD), yet a systematic method for doing so remains unknown. In this work, we propose DSV (Discordance and Separability Validation), an unsupervised validation loss to select high-performing detection models with effective augmentation HPs. DSV captures the alignment between an augmentation function and the anomaly-generating mechanism with surrogate losses, which approximate the discordance and separability of test data, respectively. As a result, the evaluation via DSV leads to selecting an effective SSAD model exhibiting better alignment, which results in high detection accuracy. We theoretically derive the degree of approximation conducted by the surrogate losses and empirically show that DSV outperforms a wide range of baselines on 21 real-world tasks.
引用
收藏
页码:254 / 269
页数:16
相关论文
共 29 条
  • [1] Baevski A, 2022, PR MACH LEARN RES
  • [2] Bergman L., 2020, INT C LEARN REPR
  • [3] MVTec AD - A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection
    Bergmann, Paul
    Fauser, Michael
    Sattlegger, David
    Steger, Carsten
    [J]. 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019), 2019, : 9584 - 9592
  • [4] An Empirical Study of Training Self-Supervised Vision Transformers
    Chen, Xinlei
    Xie, Saining
    He, Kaiming
    [J]. 2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 9620 - 9629
  • [5] DeVries T, 2017, Arxiv, DOI arXiv:1708.04552
  • [6] ProtTrans: Toward Understanding the Language of Life Through Self-Supervised Learning
    Elnaggar, Ahmed
    Heinzinger, Michael
    Dallago, Christian
    Rehawi, Ghalia
    Wang, Yu
    Jones, Llion
    Gibbs, Tom
    Feher, Tamas
    Angerer, Christoph
    Steinegger, Martin
    Bhowmik, Debsindhu
    Rost, Burkhard
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2022, 44 (10) : 7112 - 7127
  • [7] Golan I., 2018, NEURIPS
  • [8] Groggel D.J., 2000, Technometrics, V42, P317
  • [9] Deep Residual Learning for Image Recognition
    He, Kaiming
    Zhang, Xiangyu
    Ren, Shaoqing
    Sun, Jian
    [J]. 2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2016, : 770 - 778
  • [10] Jezek S., 2021, ICUMT