mlsanityData to 7 Oct 2026

Rethinking Weak Supervision in Anomaly Detection: A Comprehensive Benchmark

arXiv 2605.26068 ↗First posted 25 May 2026cs.LG cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Weakly supervised anomaly detection (WSAD) has developed in three primary directions: incomplete, inexact, and inaccurate supervision. However, these directions remain isolated, lacking a unified framework to assess whether they address unique challenges or share fundamental mechanisms. This paper introduces WSADBench, the first benchmark that unifies evaluation across distinct weakly supervised scenarios, benchmarking diverse approaches from specialized WSAD methods to advanced tabular foundation models. WSADBench establishes standardized protocols to evaluate 36 algorithms across 4 modalities by systematically varying label quantity, granularity, and quality, revealing the performance boundaries of various methods. Based on over 700K experiments, WSADBench reveals four critical insights: (i) Strong intrinsic correlations exist between these weak supervision scenarios, challenging the isolation of current research directions. (ii) Specialized WSAD algorithms excel only in extreme label-scarcity regimes but are quickly dominated by tabular foundation models and general classification methods as supervision increases or in OOD scenarios. (iii) Unlabeled data shows inconsistent utility across settings, with marginal gains compared to label refinement. (iv) Models exhibit asymmetric sensitivity to different types of label noise. We release WSADBench as an open-source benchmark with code and datasets to facilitate future WSAD research: https://github.com/SUFE-AILAB/WSADBench.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 42 of its 65 references that are in the map. First posted on arXiv 25 May 2026.

Builds on

42 of 65 references.

References · 42

  1. 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,120
  2. 2017On Calibration of Modern Neural Networkscited by 4,860
  3. 2013Auto-Encoding Variational Bayescited by 3,980
  4. 2017Quo Vadis, Action Recognition? A New Model and the Kinetics Datasetcited by 3,259
  5. 2018SlowFast Networks for Video Recognitioncited by 1,862
  6. 2021Revisiting Deep Learning Models for Tabular Datacited by 725
  7. 2018Real-world Anomaly Detection in Surveillance Videoscited by 570
  8. 2019Confident Learning: Estimating Uncertainty in Dataset Labelscited by 411
  9. 2018GANomaly: Semi-Supervised Anomaly Detection via Adversarial Trainingcited by 409
  10. 2017Future Frame Prediction for Anomaly Detection -- A New Baselinecited by 391
  11. 2019Deep Semi-Supervised Anomaly Detectioncited by 290
  12. 2022OpenOOD: Benchmarking Generalized Out-of-Distribution Detectioncited by 267
30 more
  1. 2020Not only Look, but also Listen: Learning Multimodal Violence Detection under Weak Supervisioncited by 235
  2. 2019PyOD: A Python Toolbox for Scalable Outlier Detectioncited by 221
  3. 2019Graph Convolutional Label Noise Cleaner: Train a Plug-and-play Action Classifier for Anomaly Detectioncited by 171
  4. 2022ECOD: Unsupervised Outlier Detection Using Empirical Cumulative Distribution Functionscited by 153
  5. 2019Deep Anomaly Detection with Deviation Networkscited by 152
  6. 2020DROCC: Deep Robust One-Class Classificationcited by 121
  7. 2023VadCLIP: Adapting Vision-Language Models for Weakly Supervised Video Anomaly Detectioncited by 121
  8. 2026TabICLv2: A better, faster, scalable, and open tabular foundation modelcited by 112
  9. 2024TabM: Advancing Tabular Deep Learning with Parameter-Efficient Ensemblingcited by 105
  10. 2021WRENCH: A Comprehensive Benchmark for Weak Supervisioncited by 95
  11. 2023Dual Memory Units with Uncertainty Regulation for Weakly Supervised Video Anomaly Detectioncited by 88
  12. 2022MGFN: Magnitude-Contrastive Glance-and-Focus Network for Weakly-Supervised Video Anomaly Detectioncited by 87
  13. 2023BMAD: Benchmarks for Medical Anomaly Detectioncited by 81
  14. 2018Learning Representations of Ultrahigh-dimensional Data for Random Distance-based Outlier Detectioncited by 79
  15. 2023TabR: Tabular Deep Learning Meets Nearest Neighbors in 2023cited by 73
  16. 2025Mitra: Mixed Synthetic Priors for Enhancing Tabular Foundation Modelscited by 67
  17. 2025LimiX: Unleashing Structured-Data Modeling Capability for Generalist Intelligencecited by 65
  18. 2019Deep Weakly-supervised Anomaly Detectioncited by 59
  19. 2021Weakly Supervised Video Anomaly Detection via Center-guided Discriminative Learningcited by 44
  20. 2019XGBOD: Improving Supervised Outlier Detection with Unsupervised Representation Learningcited by 42
  21. 2021Feature Encoding with AutoEncoders for Weakly-supervised Anomaly Detectioncited by 42
  22. 2021LUNAR: Unifying Local Outlier Detection Methods via Graph Neural Networkscited by 42
  23. 2024Video Anomaly Detection in 10 Years: A Survey and Outlookcited by 18
  24. 2024NLP-ADBench: NLP Anomaly Detection Benchmarkcited by 17
  25. 2022ADMoE: Anomaly Detection with Mixture-of-Experts from Noisy Labelscited by 15
  26. 2022Anomaly Detection by Leveraging Incomplete Anomalous Knowledge with Anomaly-Aware Bidirectional GANscited by 11
  27. 2022AnoShift: A Distribution Shift Benchmark for Unsupervised Anomaly Detectioncited by 11
  28. 2023RoSAS: Deep Semi-Supervised Anomaly Detection with Contamination-Resilient Continuous Supervisioncited by 11
  29. 2025Diffusion-Scheduled Denoising Autoencoders for Anomaly Detection in Tabular Datacited by 3
  30. 2025Text-ADBench: Text Anomaly Detection Benchmark Based on LLM Embeddingscited by 1

Built on by

Not cited yet.