Recent advancements in anomaly detection are increasingly focused on enhancing model robustness and interpretability across diverse applications, from industrial inspections to financial monitoring. A notable trend is the integration of physics-informed frameworks and multimodal approaches, which allow for a deeper understanding of dynamic anomalies and the contextual relationships within data. For instance, new methods are leveraging vision-language models to incorporate physical priors, significantly improving detection accuracy in complex scenarios. Zero-shot anomaly detection techniques are also evolving, with models like VisualAD and MoECLIP demonstrating superior performance without relying on extensive labeled datasets. Additionally, generative frameworks are being employed to tackle high-dimensional data challenges, providing interpretable outputs that facilitate actionable insights. The field is moving towards solutions that not only enhance detection capabilities but also address fairness and interpretability, particularly in critical sectors like power management, where operational continuity is paramount. These developments suggest a maturation of anomaly detection techniques, making them more applicable and reliable in real-world settings.
Zero-shot anomaly detection (ZSAD) requires detecting and localizing anomalies without access to target-class anomaly samples. Mainstream methods rely on vision-language models (VLMs) such as CLIP: th...
We propose Bijective Universal Scene-Specific Anomalous Relationship Detection (BUSSARD), a normalizing flow-based model for detecting anomalous relations in scene graphs, generated from images. Our w...
Vision-Language Models (VLMs) demonstrate strong general-purpose reasoning but remain limited in physics-grounded anomaly detection, where causal understanding of dynamics is essential. Existing VLMs,...
Quantum machine learning offers the ability to capture complex correlations in high-dimensional feature spaces, crucial for the challenge of detecting beyond the Standard Model physics in collider eve...
Detecting structural instability and anomalies in high-dimensional financial time series is challenging due to complex temporal dependence and evolving cross-sectional structure. We propose ReGEN-TAD,...
Vision-language models have recently shown strong generalization in zero-shot anomaly detection (ZSAD), enabling the detection of unseen anomalies without task-specific supervision. However, existing ...
Reliable anomaly detection in distributed power plant monitoring systems is essential for ensuring operational continuity and reducing maintenance costs, particularly in regions where telecom operator...
Unsupervised Continuous Anomaly Detection (UCAD) is gaining attention for effectively addressing the catastrophic forgetting and heavy computational burden issues in traditional Unsupervised Anomaly D...
Anomaly detection is nowadays increasingly used in industrial applications and processes. One of the main fields of the appliance is the visual inspection for surface anomaly detection, which aims to ...
Most real-world IoT data analysis tasks, such as clustering and anomaly event detection, are unsupervised and highly susceptible to the presence of outliers. In addition to sporadic scattered outliers...