Latent Anomaly Knowledge Excavation: Unveiling Sparse Sensitive Neurons in Vision-Language Models
This work addresses the challenge of interpretable anomaly detection for users of vision-language models, offering a novel paradigm shift from external adaptation to internal activation.
The paper tackles the problem of understanding and improving anomaly detection in vision-language models by proposing that anomaly knowledge is already embedded but latent in pre-trained models, and it introduces LAKE, a training-free framework that identifies sparse sensitive neurons to achieve state-of-the-art performance on industrial benchmarks.
Large-scale vision-language models (VLMs) exhibit remarkable zero-shot capabilities, yet the internal mechanisms driving their anomaly detection (AD) performance remain poorly understood. Current methods predominantly treat VLMs as black-box feature extractors, assuming that anomaly-specific knowledge must be acquired through external adapters or memory banks. In this paper, we challenge this assumption by arguing that anomaly knowledge is intrinsically embedded within pre-trained models but remains latent and under-activated. We hypothesize that this knowledge is concentrated within a sparse subset of anomaly-sensitive neurons. To validate this, we propose latent anomaly knowledge excavation (LAKE), a training-free framework that identifies and elicits these critical neuronal signals using only a minimal set of normal samples. By isolating these sensitive neurons, LAKE constructs a highly compact normality representation that integrates visual structural deviations with cross-modal semantic activations. Extensive experiments on industrial AD benchmarks demonstrate that LAKE achieves state-of-the-art performance while providing intrinsic, neuron-level interpretability. Ultimately, our work advocates for a paradigm shift: redefining anomaly detection as the targeted activation of latent pre-trained knowledge rather than the acquisition of a downstream task.