LAKE identifies sparse anomaly-sensitive neurons in pre-trained VLMs using minimal normal samples to build compact normality representations and achieve SOTA anomaly detection with neuron-level interpretability.
Towards zero- shot anomaly detection and reasoning with multimodal large language models,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Latent Anomaly Knowledge Excavation: Unveiling Sparse Sensitive Neurons in Vision-Language Models
LAKE identifies sparse anomaly-sensitive neurons in pre-trained VLMs using minimal normal samples to build compact normality representations and achieve SOTA anomaly detection with neuron-level interpretability.