Explainable Deep One-Class Classification

TL;DR

This paper introduces FCDD, an explainable deep one-class classification method that produces heatmap explanations for anomaly detection, achieving state-of-the-art results and revealing model vulnerabilities.

Contribution

FCDD is the first to generate explanation heatmaps directly from the feature space in deep one-class classification, improving interpretability and detection performance.

Findings

FCDD achieves state-of-the-art results on MVTec-AD.

Ground-truth anomaly maps enhance detection when used in training.

Models are vulnerable to spurious features like watermarks.

Abstract

Deep one-class classification variants for anomaly detection learn a mapping that concentrates nominal samples in feature space causing anomalies to be mapped away. Because this transformation is highly non-linear, finding interpretations poses a significant challenge. In this paper we present an explainable deep one-class classification method, Fully Convolutional Data Description (FCDD), where the mapped samples are themselves also an explanation heatmap. FCDD yields competitive detection performance and provides reasonable explanations on common anomaly detection benchmarks with CIFAR-10 and ImageNet. On MVTec-AD, a recent manufacturing dataset offering ground-truth anomaly maps, FCDD sets a new state of the art in the unsupervised setting. Our method can incorporate ground-truth anomaly maps during training and using even a few of these (~5) improves performance significantly. Finally, using FCDD's explanations we demonstrate the vulnerability of deep one-class classification models to spurious image features such as image watermarks.