Set Features for Anomaly Detection
It addresses a specific limitation in anomaly detection for scenarios like logical anomalies in images or time series, offering a novel approach but is incremental in improving over prior methods.
The paper tackles the problem of detecting anomalies that arise from unusual combinations of normal elements, which existing segmentation-based methods fail to handle well, and achieves state-of-the-art performance in image-level logical anomaly detection and sequence-level time series anomaly detection.
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.