Advancements in Anomaly Detection

The field of anomaly detection is moving towards more general and robust methods that can handle diverse domains and tasks. Researchers are exploring new approaches to address the limitations of traditional methods, such as class-agnostic anomaly detection and temporal-topological scattering mechanisms. These innovative methods aim to improve the detection of anomalies in complex data, including time series and spatiotemporal data. Noteworthy papers in this area include ResAD++, which proposes a residual feature learning approach for class-agnostic anomaly detection, and ScatterAD, which introduces a temporal-topological scattering mechanism for time series anomaly detection. Additionally, the Normal-Abnormal Guided Generalist Anomaly Detection approach leverages both normal and anomalous samples as references to guide anomaly detection across diverse domains.

Sources

ResAD++: Towards Class Agnostic Anomaly Detection via Residual Feature Learning

ScatterAD: Temporal-Topological Scattering Mechanism for Time Series Anomaly Detection

Generalist Multi-Class Anomaly Detection via Distillation to Two Heterogeneous Student Networks

Anomaly detection for generic failure monitoring in robotic assembly, screwing and manipulation

Robust Spatiotemporally Contiguous Anomaly Detection Using Tensor Decomposition

Normal-Abnormal Guided Generalist Anomaly Detection

Built with on top of