# Mean-Shifted Contrastive Loss for Anomaly Detection

7 Jun 2021  ·  , ·

Deep anomaly detection methods learn representations that separate between normal and anomalous images. Although self-supervised representation learning is commonly used, small dataset sizes limit its effectiveness. It was previously shown that utilizing external, generic datasets (e.g. ImageNet classification) can significantly improve anomaly detection performance. One approach is outlier exposure, which fails when the external datasets do not resemble the anomalies. We take the approach of transferring representations pre-trained on external datasets for anomaly detection. Anomaly detection performance can be significantly improved by fine-tuning the pre-trained representations on the normal training images. In this paper, we first demonstrate and analyze that contrastive learning, the most popular self-supervised learning paradigm cannot be naively applied to pre-trained features. The reason is that pre-trained feature initialization causes poor conditioning for standard contrastive objectives, resulting in bad optimization dynamics. Based on our analysis, we provide a modified contrastive objective, the Mean-Shifted Contrastive Loss. Our method is highly effective and achieves a new state-of-the-art anomaly detection performance including $98.6\%$ ROC-AUC on the CIFAR-10 dataset.

PDF Abstract

## Results from the Paper Edit

Ranked #3 on Anomaly Detection on One-class CIFAR-100 (using extra training data)

Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Anomaly Detection Anomaly Detection on Unlabeled CIFAR-10 vs LSUN (Fix) MeanShifted ROC-AUC 92.6 # 5
Network ResNet-152 # 1
Anomaly Detection MVTec AD Mean-Shifted Contrastive Loss Detection AUROC 87.2 # 54
Anomaly Detection One-class CIFAR-10 Mean-Shifted Contrastive Loss AUROC 98.6 # 4
Anomaly Detection One-class CIFAR-100 Mean-Shifted Contrastive Loss AUROC 96.5 # 3
Anomaly Detection Unlabeled CIFAR-10 vs CIFAR-100 MeanShifted AUROC 90.0 # 5
Network ResNet-152 # 1