A multi-task scheme with synthetic anomalies from graph perturbations and two-phase training learns robust features for weakly supervised graph anomaly detection, outperforming competitors on public datasets.
arXiv preprint arXiv:2409.09957
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
verdicts
UNVERDICTED 4roles
background 2polarities
background 2representative citing papers
RTTAD improves unsupervised tabular anomaly detection by combining collaborative dual-task learning during training with selective, risk-aware test-time contrastive learning that avoids anomaly contamination.
NK-GAD improves unsupervised graph anomaly detection on heterophilic graphs by combining a joint encoder for similar and dissimilar neighbors, neighbor reconstruction, center aggregation, and dual decoders, yielding an average 3.29% AUC gain across seven datasets.
BlindGuard introduces an unsupervised hierarchical agent encoder plus corruption-guided contrastive detector that identifies malicious agents in LLM-based multi-agent systems without any attack labels or prior knowledge of malicious behaviors.
citing papers explorer
-
Learning Feature Encoder with Synthetic Anomalies for Weakly Supervised Graph Anomaly Detection
A multi-task scheme with synthetic anomalies from graph perturbations and two-phase training learns robust features for weakly supervised graph anomaly detection, outperforming competitors on public datasets.
-
When Normality Shifts: Risk-Aware Test-Time Adaptation for Unsupervised Tabular Anomaly Detection
RTTAD improves unsupervised tabular anomaly detection by combining collaborative dual-task learning during training with selective, risk-aware test-time contrastive learning that avoids anomaly contamination.
-
NK-GAD: Neighbor Knowledge-Enhanced Unsupervised Graph Anomaly Detection
NK-GAD improves unsupervised graph anomaly detection on heterophilic graphs by combining a joint encoder for similar and dissimilar neighbors, neighbor reconstruction, center aggregation, and dual decoders, yielding an average 3.29% AUC gain across seven datasets.
-
BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks
BlindGuard introduces an unsupervised hierarchical agent encoder plus corruption-guided contrastive detector that identifies malicious agents in LLM-based multi-agent systems without any attack labels or prior knowledge of malicious behaviors.