mlsanityData to 7 Oct 2026

Elsa: Energy-based learning for semi-supervised anomaly detection

arXiv 2103.15296 ↗First posted 29 Mar 2021cs.CV cs.LGcited by 6

Not on the map. Why?

Abstract (arXiv)

Anomaly detection aims at identifying deviant instances from the normal data distribution. Many advances have been made in the field, including the innovative use of unsupervised contrastive learning. However, existing methods generally assume clean training data and are limited when the data contain unknown anomalies. This paper presents Elsa, a novel semi-supervised anomaly detection approach that unifies the concept of energy-based models with unsupervised contrastive learning. Elsa instills robustness against any data contamination by a carefully designed fine-tuning step based on the new energy function that forces the normal data to be divided into classes of prototypes. Experiments on multiple contamination scenarios show the proposed model achieves SOTA performance. Extensive analyses also verify the contribution of each component in the proposed model. Beyond the experiments, we also offer a theoretical interpretation of why contrastive learning alone cannot detect anomalies under data contamination.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 33 of its 48 references that are in the map. First posted on arXiv 29 Mar 2021.

Builds on

33 of 48 references.

References · 33

  1. 2020A Simple Framework for Contrastive Learning of Visual Representationscited by 11,979
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  3. 2019Momentum Contrast for Unsupervised Visual Representation Learningcited by 7,482
  4. 2019RandAugment: Practical automated data augmentation with a reduced search spacecited by 2,261
  5. 2020FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidencecited by 2,080
  6. 2017Improved Regularization of Convolutional Neural Networks with Cutoutcited by 1,623
  7. 2020Understanding Contrastive Representation Learning through Alignment and Uniformity on the Hyperspherecited by 1,393
  8. 2017Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networkscited by 1,325
  9. 2020Energy-based Out-of-distribution Detectioncited by 1,296
  10. 2019Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shiftcited by 1,274
  11. 2017Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discoverycited by 566
  12. 2019Using Self-Supervised Learning Can Improve Model Robustness and Uncertaintycited by 511
21 more
  1. 2014Semi-Supervised Learning with Deep Generative Modelscited by 494
  2. 2020CSI: Novelty Detection via Contrastive Learning on Distributionally Shifted Instancescited by 441
  3. 2018GANomaly: Semi-Supervised Anomaly Detection via Adversarial Trainingcited by 409
  4. 2019Your Classifier is Secretly an Energy Based Model and You Should Treat it Like Onecited by 402
  5. 2019Deep Semi-Supervised Anomaly Detectioncited by 290
  6. 2020Identifying Mislabeled Data using the Area Under the Margin Rankingcited by 259
  7. 2020Classification-Based Anomaly Detection for General Datacited by 210
  8. 2019OCGAN: One-class Novelty Detection Using GANs with Constrained Latent Representationscited by 199
  9. 2018Latent Space Autoregression for Novelty Detectioncited by 149
  10. 2020Intriguing Properties of Contrastive Lossescited by 129
  11. 2018Adversarially Learned Anomaly Detectioncited by 115
  12. 2020Old is Gold: Redefining the Adversarially Learned One-Class Classifier Training Paradigmcited by 115
  13. 2020Learning Latent Space Energy-Based Prior Modelcited by 104
  14. 2019Flow Contrastive Estimation of Energy-Based Modelscited by 75
  15. 2014Toward Supervised Anomaly Detectioncited by 73
  16. 2020Backpropagated Gradient Representations for Anomaly Detectioncited by 54
  17. 2020Improving Unsupervised Image Clustering With Robust Learningcited by 51
  18. 2018One-Class Adversarial Nets for Fraud Detectioncited by 33
  19. 2019Fence GAN: Towards Better Anomaly Detectioncited by 32
  20. 2020Joint Training of Variational Auto-Encoder and Latent Energy-Based Modelcited by 30
  21. 2019Novelty Detection Via Blurringcited by 24

Built on by

Cited by year

Citing papers per year: 2021: 1; 2022: 3; 2023: 0; 2024: 1; 2025: 1.

6 papers.

  1. 2025Quantum Federated Learning: A Comprehensive Surveycited by 7
  2. 2021Efficient Anomaly Detection Using Self-Supervised Multi-Cue Taskscited by 6
  3. 2022Hierarchical Semi-Supervised Contrastive Learning for Contamination-Resistant Anomaly Detectioncited by 5
  4. 2022No Shifted Augmentations (NSA): compact distributions for robust self-supervised Anomaly Detectioncited by 3
  5. 2024Adaptive Graph Learning from Spatial Information for Surgical Workflow Anticipationcited by 2
  6. 2022Anomaly Detection via Multi-Scale Contrasted Memorycited by 0