Abstract:Graph anomaly detection (GAD) is a critical task in graph machine learning, with the primary objective of identifying anomalous nodes that deviate significantly from the majority. This task is widely applied in various real-world scenarios, including fraud detection and social network analysis. However, existing GAD methods still face two major challenges: (1) They are often limited to detecting anomalies in single-type interaction graphs and struggle with multiple interaction types in multiplex heterogeneous graphs; (2) In unsupervised scenarios, selecting appropriate anomaly score thresholds remains a significant challenge for accurate anomaly detection. To address the above challenges, we propose a novel Unsupervised Multiplex Graph Anomaly Detection method, named UMGAD. We first learn multi-relational correlations among nodes in multiplex heterogeneous graphs and capture anomaly information during node attribute and structure reconstruction through graph-masked autoencoder (GMAE). Then, to further weaken the influence of noise and redundant information on abnormal information extraction, we generate attribute-level and subgraph-level augmented-view graphs respectively, and perform attribute and structure reconstruction through GMAE. Finally, We learn to optimize node attributes and structural features through contrastive learning between original-view and augmented-view graphs to improve the model's ability to capture anomalies. Meanwhile, we also propose a new anomaly score threshold selection strategy, which allows the model to be independent of the ground truth in real unsupervised scenarios. Extensive experiments on four datasets show that our \model significantly outperforms state-of-the-art methods, achieving average improvements of 13.48% in AUC and 11.68% in Macro-F1 across all datasets.

Improving Generalizability of Graph Anomaly Detection Models Via Data Augmentation

Improving Generalizability of Graph Anomaly Detection Models via Data Augmentation

DAGAD: Data Augmentation for Graph Anomaly Detection

Generative Semi-supervised Graph Anomaly Detection

Counterfactual Data Augmentation with Denoising Diffusion for Graph Anomaly Detection

Accurate Anomaly Detection Leveraging Knowledge-enhanced GAT

GADY: Unsupervised Anomaly Detection on Dynamic Graphs

Generative and Contrastive Self-Supervised Learning for Graph Anomaly Detection

Imbalanced Graph-Level Anomaly Detection via Counterfactual Augmentation and Feature Learning

Data Augmentation on Graphs: A Technical Survey

Graph Pre-Training Models Are Strong Anomaly Detectors

Graph Anomaly Detection Via Multi-Scale Contrastive Learning Networks with Augmented View.

A Universal Adaptive Algorithm for Graph Anomaly Detection

LabelGen: An Anomaly Label Generative Framework for Enhanced Graph Anomaly Detection

DOPING: Generative Data Augmentation for Unsupervised Anomaly Detection with GAN

UMGAD: Unsupervised Multiplex Graph Anomaly Detection

Mul-GAD: a semi-supervised graph anomaly detection framework via aggregating multi-view information

AutoAUG : Automatic Data Augmentation for Graph Neural Networks.

Graph Anomaly Detection at Group Level: A Topology Pattern Enhanced Unsupervised Approach

Towards data augmentation in graph neural network: An overview and evaluation

Improving Graph Convolutional Network with Learnable Edge Weights and Edge-Node Co-Embedding for Graph Anomaly Detection