🤖 AI Summary
This work addresses the challenging problem of detecting logical anomalies—such as semantic inconsistencies in images or temporal state mismatches in sequences—arising from abnormal combinations of otherwise normal elements. Moving beyond conventional local anomaly detection paradigms, we propose a global distribution modeling approach based on “set features”: each sample is represented as the joint distribution over its constituent elements, rather than as isolated points or segmentation outputs. Our method employs a lightweight, training-free, unsupervised framework comprising fixed-feature extraction followed by kernel density estimation (KDE). The key innovation lies in the first formal definition and construction of set features, enabling explicit modeling of combinatorial structural anomalies. Evaluated on both image-level logical anomaly detection and sequence-level time-series anomaly detection, our approach achieves significant improvements over current state-of-the-art methods, delivering higher accuracy and superior cross-domain generalization capability.
📝 Abstract
This paper proposes to use set features for detecting anomalies in samples that consist of unusual combinations of normal elements. Many leading methods discover anomalies by detecting an unusual part of a sample. For example, state-of-the-art segmentation-based approaches, first classify each element of the sample (e.g., image patch) as normal or anomalous and then classify the entire sample as anomalous if it contains anomalous elements. However, such approaches do not extend well to scenarios where the anomalies are expressed by an unusual combination of normal elements. In this paper, we overcome this limitation by proposing set features that model each sample by the distribution of its elements. We compute the anomaly score of each sample using a simple density estimation method, using fixed features. Our approach outperforms the previous state-of-the-art in image-level logical anomaly detection and sequence-level time series anomaly detection.
