A Feature Shuffling and Restoration Strategy for Universal Unsupervised Anomaly Detection

ArXi:2603.22861v1 Announce Type: new Unsupervised anomaly detection is vital in industrial fields, with reconstruction-based methods favored for their simplicity and effectiveness. However, reconstruction methods often encounter an identical shortcut issue, where both normal and anomalous regions can be well reconstructed and fail to identify outliers. The severity of this problem increases with the complexity of the normal data distribution.