AC2L-GAD: Active Counterfactual Contrastive Learning for Graph Anomaly Detection

2026-01-29

AI Summary

This work addresses the challenges of graph anomaly detection under extreme class imbalance and severe label scarcity, where existing contrastive learning approaches often suffer from semantic inconsistency due to random augmentations and non-discriminative negative sampling. To overcome these limitations, the authors propose an active counterfactual contrastive learning framework that uniquely integrates active learning with counterfactual reasoning. By leveraging an information-theoretic acquisition function, the method actively selects informative subgraphs and generates semantics-preserving positive samples along with normal negative samples, thereby constructing challenging yet semantically coherent contrastive pairs for effective hard negative mining. Evaluated on nine benchmark datasets—including the GADBench financial transaction graph—the approach achieves state-of-the-art or competitive performance, particularly excelling in scenarios involving complex attribute-structure interactions, while reducing computational overhead by approximately 65%.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Graph anomaly detection aims to identify abnormal patterns in networks, but faces significant challenges from label scarcity and extreme class imbalance. While graph contrastive learning offers a promising unsupervised solution, existing methods suffer from two critical limitations: random augmentations break semantic consistency in positive pairs, while naive negative sampling produces trivial, uninformative contrasts. We propose AC2L-GAD, an Active Counterfactual Contrastive Learning framework that addresses both limitations through principled counterfactual reasoning. By combining information-theoretic active selection with counterfactual generation, our approach identifies structurally complex nodes and generates anomaly-preserving positive augmentations alongside normal negative counterparts that provide hard contrasts, while restricting expensive counterfactual generation to a strategically selected subset. This design reduces computational overhead by approximately 65% compared to full-graph counterfactual generation while maintaining detection quality. Experiments on nine benchmark datasets, including real-world financial transaction graphs from GADBench, show that AC2L-GAD achieves competitive or superior performance compared to state-of-the-art baselines, with notable gains in datasets where anomalies exhibit complex attribute-structure interactions.

Problem

• graph anomaly detection
• label scarcity
• class imbalance
• contrastive learning
• counterfactual reasoning

Innovation

• Counterfactual Generation
• Active Sampling
• Graph Contrastive Learning
• Anomaly Detection
• Semantic Consistency