DCOR: Anomaly Detection in Attributed Networks via Dual Contrastive Learning Reconstruction

TL;DR

This work tackles anomaly detection in attributed networks by integrating reconstruction-based signaling with contrastive learning in a Graph Neural Network framework. DCOR uses a dual autoencoder to reconstruct both the adjacency matrix $\hat{A}$ and the attribute matrix $\hat{X}$ and then applies a contrastive objective between reconstructions from the original and augmented graphs, optimizing a total loss $\mathcal{L}_{total} = \lambda_{\text{rec}} \mathcal{L}_{\text{rec}} + \lambda_{\text{sc}} \mathcal{L}^{sc}$. The key contributions are the dual reconstruction mechanism, graph data augmentations (node-level and structural), and a contrastive loss that specifically exploits reconstruction differences to better identify subtle structural and attribute anomalies. Empirical results on Flickr, Amazon, Enron, and Facebook show DCOR outperforms several baselines, with ablation studies demonstrating the value of reconstruction-focused contrastive learning. The approach offers a robust, end-to-end framework for detecting anomalies in complex attributed networks with potential applicability across finance, security, and network fault domains.

Abstract

Anomaly detection using a network-based approach is one of the most efficient ways to identify abnormal events such as fraud, security breaches, and system faults in a variety of applied domains. While most of the earlier works address the complex nature of graph-structured data and predefined anomalies, the impact of data attributes and emerging anomalies are often neglected. This paper introduces DCOR, a novel approach on attributed networks that integrates reconstruction-based anomaly detection with Contrastive Learning. Utilizing a Graph Neural Network (GNN) framework, DCOR contrasts the reconstructed adjacency and feature matrices from both the original and augmented graphs to detect subtle anomalies. We employed comprehensive experimental studies on benchmark datasets through standard evaluation measures. The results show that DCOR significantly outperforms state-of-the-art methods. Obtained results demonstrate the efficacy of proposed approach in attributed networks with the potential of uncovering new patterns of anomalies.