Towards Anomaly-Aware Pre-Training and Fine-Tuning for Graph Anomaly Detection

2025/04/19 by Yunhui Liu, Jiashun Cheng, Liu, Yunhui +17 · 1 citation
Computer Science · #Advanced Graph Neural Networks #Anomaly Detection Techniques and Applications #FOS: Computer and information sciences #Machine Learning (cs.LG) #Software System Performance and Reliability

paper · pdf · doi:10.48550/arxiv.2504.14250

openalex publication_date 2025/04/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Graph anomaly detection (GAD) has garnered increasing attention in recent years, yet remains challenging due to two key factors: (1) label scarcity stemming from the high cost of annotations and (2) homophily disparity at node and class levels. In this paper, we introduce Anomaly-Aware Pre-Training and Fine-Tuning (APF), a targeted and effective framework to mitigate the above challenges in GAD. In the pre-training stage, APF incorporates node-specific subgraphs selected via the Rayleigh Quotient, a label-free anomaly metric, into the learning objective to enhance anomaly awareness. It further introduces two learnable spectral polynomial filters to jointly learn dual representations that capture both general semantics and subtle anomaly cues. During fine-tuning, a gated fusion mechanism adaptively integrates pre-trained representations across nodes and dimensions, while an anomaly-aware regularization loss encourages abnormal nodes to preserve more anomaly-relevant information. Furthermore, we theoretically show that APF tends to achieve linear separability under mild conditions. Comprehensive experiments on 10 benchmark datasets validate the superior performance of APF in comparison to state-of-the-art baselines.

Citations

• AnomalyGFM: Graph Foundation Model for Zero/Few-shot Anomaly Detection
• SpaceGNN: Multi-Space Graph Neural Network for Node Anomaly Detection with Extremely Limited Labels
• Partitioning Message Passing for Graph Fraud Detection
• UniGAD: Unifying Multi-level Graph Anomaly Detection
• Graph Neural Networks for Financial Fraud Detection: A Review
• Graph Pre-Training Models Are Strong Anomaly Detectors
• Zero-shot Generalist Graph Anomaly Detection with Unified Neighborhood Prompts
• DiffGAD: A Diffusion-based Unsupervised Graph Anomaly Detector
• Deep Graph Anomaly Detection: A Survey and New Perspectives
• Bootstrap Latents of Nodes and Neighbors for Graph Self-Supervised Learning
• Investigating Self-Supervised Methods for Label-Efficient Learning
• Node-wise Filtering in Graph Neural Networks: A Mixture of Experts Approach
• ARC: A Generalist Graph Anomaly Detector with In-Context Learning
• Generative Semi-supervised Graph Anomaly Detection
• A Survey of Data-Efficient Graph Learning
• Understanding Heterophily for Graph Neural Networks
• Open-Set Graph Anomaly Detection via Normal Structure Regularisation
• Rayleigh Quotient Graph Neural Networks for Graph-level Anomaly Detection
• GADBench: Revisiting and Benchmarking Supervised Graph Anomaly Detection
• Demystifying Structural Disparity in Graph Neural Networks: Can One Size Fit All?
• Truncated Affinity Maximization: One-class Homophily Modeling for Graph Anomaly Detection
• Rethinking Semi-supervised Learning with Language Models
• A critical look at the evaluation of GNNs under heterophily: Are we really making progress?
• Cross-Domain Graph Anomaly Detection via Anomaly-aware Contrastive Alignment
• Revisiting Heterophily For Graph Neural Networks
• DGraph: A Large-Scale Financial Dataset for Graph Anomaly Detection
• Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning
• BOND: Benchmarking Unsupervised Outlier Node Detection on Static Attributed Graphs
• Rethinking Graph Neural Networks for Anomaly Detection
• EvenNet: Ignoring Odd-Hop Neighbors Improves Robustness of Graph Neural Networks
• How Powerful are Spectral Graph Neural Networks
• GraphMAE: Self-Supervised Masked Graph Autoencoders
• Convolutional Neural Networks on Graphs with Chebyshev Approximation, Revisited
• From Canonical Correlation Analysis to Self-supervised Graph Neural Networks
• BernNet: Learning Arbitrary Graph Spectral Filters via Bernstein Approximation
• Is Homophily a Necessity for Graph Neural Networks?
• AdaGNN: Graph Neural Networks with Adaptive Frequency Response Filter
• Anomaly Detection on Attributed Networks via Contrastive Self-Supervised Learning
• Few-shot Network Anomaly Detection via Cross-network Meta-learning
• Graph Convolution for Semi-Supervised Classification: Improved Linear Separability and Out-of-Distribution Generalization
• Large-Scale Representation Learning on Graphs via Bootstrapping
• In Defense of Pseudo-Labeling: An Uncertainty-Aware Pseudo-label Selection Framework for Semi-Supervised Learning
• Beyond Low-frequency Information in Graph Convolutional Networks
• Graph Contrastive Learning with Adaptive Augmentation
• Enhancing Graph Neural Network-based Fraud Detectors against Camouflaged Fraudsters
• Big Self-Supervised Models are Strong Semi-Supervised Learners
• Deep Graph Contrastive Representation Learning
• Deep Weakly-supervised Anomaly Detection
• Spam Review Detection with Graph Convolutional Networks
• Anti-Money Laundering in Bitcoin: Experimenting with Graph Convolutional Networks for Financial Forensics
• Predicting Dynamic Embedding Trajectory in Temporal Interaction Networks
• How Powerful are Graph Neural Networks?
• Deep Graph Infomax
• Learning deep representations by mutual information estimation and maximization
• Contextual Stochastic Block Models
• Graph Attention Networks
• Semi-Supervised Classification with Graph Convolutional Networks
• XGBoost: A Scalable Tree Boosting System
• Consistency of spectral clustering in stochastic block models
• From Amateurs to Connoisseurs: Modeling the Evolution of User Expertise through Online Reviews
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

Cited by

• Semi-supervised Graph Anomaly Detection via Robust Homophily Learning

Related

• Graph Pre-Training Models Are Strong Anomaly Detectors
• ARC: A Generalist Graph Anomaly Detector with In-Context Learning
• PREM: A Simple Yet Effective Approach for Node-Level Graph Anomaly Detection
• AnomalyGFM: Graph Foundation Model for Zero/Few-shot Anomaly Detection
• FreeGAD: A Training-Free yet Effective Approach for Graph Anomaly Detection