BayPrAnoMeta: Bayesian Proto-MAML for Few-Shot Industrial Image Anomaly Detection

TL;DR

BayPrAnoMeta introduces a Bayesian approach to few-shot industrial image anomaly detection, replacing deterministic prototypes with probabilistic models to improve robustness and uncertainty estimation, especially in extreme few-shot scenarios.

Contribution

It proposes a Bayesian generalization of Proto-MAML using task-specific probabilistic normality models and extends to federated meta-learning with contrastive regularization.

Findings

Achieves significant AUROC improvements over existing methods.

Demonstrates robustness in extreme few-shot settings.

Validates effectiveness on the MVTec AD benchmark.

Abstract

Industrial image anomaly detection is a challenging problem owing to extreme class imbalance and the scarcity of labeled defective samples, particularly in few-shot settings. We propose BayPrAnoMeta, a Bayesian generalization of Proto-MAML for few-shot industrial image anomaly detection. Unlike existing Proto-MAML approaches that rely on deterministic class prototypes and distance-based adaptation, BayPrAnoMeta replaces prototypes with task-specific probabilistic normality models and performs inner-loop adaptation via a Bayesian posterior predictive likelihood. We model normal support embeddings with a Normal-Inverse-Wishart (NIW) prior, producing a Student-$t$ predictive distribution that enables uncertainty-aware, heavy-tailed anomaly scoring and is essential for robustness in extreme few-shot settings. We further extend BayPrAnoMeta to a federated meta-learning framework with supervised contrastive regularization for heterogeneous industrial clients and prove convergence to stationary points of the resulting nonconvex objective. Experiments on the MVTec AD benchmark demonstrate consistent and significant AUROC improvements over MAML, Proto-MAML, and PatchCore-based methods in few-shot anomaly detection settings.