New Chi-Square Wavelet GNN Framework Enhances Heterogeneous Graph Anomaly Detection

Researchers have developed a novel spectral Graph Neural Network (GNN) framework called ChiGAD for anomaly detection in heterogeneous networks. This framework addresses key challenges such as capturing diverse meta-path semantics, preserving high-frequency content during dimension alignment, and handling class imbalance with difficult anomaly samples. ChiGAD utilizes a Multi-Graph Chi-Square Filter, Interactive Meta-Graph Convolution, and a Contribution-Informed Cross-Entropy Loss to improve performance. Experiments show ChiGAD outperforms existing state-of-the-art models, and its variant, ChiGNN, also demonstrates effectiveness on GAD datasets. AI

IMPACT This research introduces a novel framework that could improve anomaly detection in complex, heterogeneous graph data, potentially benefiting cybersecurity and fraud detection applications.

RANK_REASON The cluster contains a research paper detailing a new GNN framework for anomaly detection. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• ChiGNN
• Chi-Square Wavelet Graph Neural Networks
• Graph Anomaly Detection
• Graph Neural Network
• Xiping Li

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →