New AI methods tackle anomaly detection with agents and graph analysis
ByPulseAugur Editorial·[12 sources]·
Researchers have developed several new methods for anomaly detection, focusing on agentic frameworks and graph-based approaches. AnomalyAgent utilizes multimodal large language models for zero- and few-shot anomaly detection, demonstrating superior reasoning capabilities over traditional similarity-based methods. Other research introduces self-designing agentic workflows for graph anomaly detection and prototype-based distillation for generalist, unsupervised detection. Additionally, a training-free Laplacian energy minimization approach reframes anomaly scoring as non-conformity, and a relational fingerprint method aims to improve generalizability in graph anomaly detection by aligning semantic features.
AI
IMPACT
Advances in anomaly detection techniques could improve robustness and efficiency in various AI applications, from cybersecurity to image analysis.
RANK_REASON
Multiple arXiv papers introducing novel research methods for anomaly detection.
Benefiting from generalizability of vision-language models (VLMs) such as CLIP, many zero-/few-shot anomaly detection (AD) approaches have achieved impressive detection performance across various datasets. Nevertheless, they require substantial training on large auxiliary dataset…
arXiv cs.AI
TIER_1English(EN)·Tairan Huang, Qiang Chen, Yili Wang, Yueyue Ma, Changlong He, Xiu Su, Yi Chen·
arXiv:2605.27470v1 Announce Type: cross Abstract: Graph anomaly detection aims to identify anomaly nodes in attributed graphs and plays an important role in real-world applications. However, existing graph anomaly detection methods still face two key challenges: 1) fixed pipeline…
arXiv cs.AI
TIER_1English(EN)·Jungwook Seo, Minjeong Kim, Younkwan Lee, Seungho Shin, Sungyong Baik·
arXiv:2605.28428v1 Announce Type: cross Abstract: Detecting subtle visual anomalies in images remains challenging, particularly when only normal samples are available a priori. Such unsupervised anomaly detection is typically solved by measuring feature similarity of a query patc…
arXiv cs.LG
TIER_1English(EN)·Yiming Xu, Zihan Chen, Zhen Peng, Song Wang, Bin Shi, Bo Dong, Chao Shen·
arXiv:2605.26857v1 Announce Type: new Abstract: Driven by the pressing demand for graph anomaly detection (GAD) in high-stakes domains, the generalist GAD paradigm, which trains a single detector transferable across new graphs, has recently gained growing attention. However, exis…
arXiv:2605.26446v1 Announce Type: cross Abstract: Graph anomaly detection (GAD) aims to identify nodes or substructures whose behavior or attributes deviate significantly from the overall pattern in graph-structured data, with critical applications in financial risk control, soci…
Driven by the pressing demand for graph anomaly detection (GAD) in high-stakes domains, the generalist GAD paradigm, which trains a single detector transferable across new graphs, has recently gained growing attention. However, existing methods often rely on scarce and costly ann…
Driven by the pressing demand for graph anomaly detection (GAD) in high-stakes domains, the generalist GAD paradigm, which trains a single detector transferable across new graphs, has recently gained growing attention. However, existing methods often rely on scarce and costly ann…
arXiv:2605.25429v1 Announce Type: new Abstract: Generalist graph anomaly detection (GAD) aims to detect anomalies on unseen graphs without graph-specific retraining. Nevertheless, existing approaches primarily focus on aligning heterogeneous features across different data domains…
arXiv:2605.30140v1 Announce Type: new Abstract: Benefiting from generalizability of vision-language models (VLMs) such as CLIP, many zero-/few-shot anomaly detection (AD) approaches have achieved impressive detection performance across various datasets. Nevertheless, they require…
Benefiting from generalizability of vision-language models (VLMs) such as CLIP, many zero-/few-shot anomaly detection (AD) approaches have achieved impressive detection performance across various datasets. Nevertheless, they require substantial training on large auxiliary dataset…
Detecting subtle visual anomalies in images remains challenging, particularly when only normal samples are available a priori. Such unsupervised anomaly detection is typically solved by measuring feature similarity of a query patch to a memory of normal patches. However, similari…
arXiv stat.ML
TIER_1English(EN)·Xudong Wang, Ziheng Sun, Chris Ding, Jicong Fan·
arXiv:2505.21285v4 Announce Type: replace-cross Abstract: This work proposes a framework LGKDE that learns kernel density estimation for graphs. The key challenge in graph density estimation lies in effectively capturing both structural patterns and semantic variations while main…
