Beyond Normal References: Discriminative Few-Shot Anomaly Detection

2026-05-22

AI Summary

This work addresses the challenge of few-shot anomaly detection, where existing methods either rely solely on normal samples or overfit to seen anomalies, limiting their generalization to unseen anomaly types. To overcome this, the authors propose IDEAL, a novel framework that jointly leverages both normal and anomalous reference samples through a two-stage architecture for generalizable anomaly representation learning. First, a Normal Variation Eraser removes irrelevant variations from normal samples; then, an Intrinsic Deviation Encoder extracts orthogonal and maximally discriminative intrinsic deviation directions. Anomaly scores are derived via a projection-based scoring mechanism. Extensive experiments across eight real-world datasets demonstrate that IDEAL significantly outperforms current state-of-the-art methods, markedly improving generalization to previously unseen anomalies.

Abstract

This paper considers a practical few-shot anomaly detection (FSAD) setting, termed discriminative FSAD, where a limited number of both normal and anomalous examples are available as references during inference. Existing FSAD methods rely on normal-only references through normality matching, ignoring the discriminative clues in anomalous references, while directly fitting both references can overfit to the seen anomalies. We introduce IDEAL, an intrinsic deviation learning framework that leverages both reference types to learn intrinsic deviation patterns characterizing generalizable abnormality as deviations from normality. IDEAL decomposes the learning process into two novel components: 1) a Normal Variation Eraser to suppress nuisance normal variations that may lead to noisy deviations from normality, thereby highlighting anomaly-relevant deviation representations; 2) an Intrinsic Deviation Encoder to decompose these denoised deviation representations into intrinsic deviation vectors capturing the most discriminative orthogonal deviation directions. At inference, IDEAL scores query-to-normal deviations preserved after projection onto the learned intrinsic deviation vectors, enabling generalization for both seen and unseen anomalies. Extensive experiments on eight real-world datasets show that IDEAL generalizes effectively to unseen anomalies and consistently outperforms existing state-of-the-art FSAD methods. Code and data will be available at \href{https://github.com/mala-lab/IDEAL}{https://github.com/mala-lab/IDEAL}.

Problem

• few-shot anomaly detection
• discriminative learning
• anomaly generalization
• normality deviation
• reference-based detection

Innovation

• few-shot anomaly detection
• intrinsic deviation learning
• normal variation eraser
• intrinsic deviation encoder
• discriminative anomaly detection