Rethinking Reconstruction-based Graph-Level Anomaly Detection: Limitations and a Simple Remedy

TL;DR

This paper identifies limitations in existing graph autoencoder-based anomaly detection methods due to a phenomenon called reconstruction flip and proposes a simple, effective solution called MUSE that leverages multifaceted error summaries to improve detection accuracy.

Contribution

The paper reveals the limitations of current reconstruction-based GLAD methods and introduces MUSE, a simple approach using multifaceted error summaries that achieves state-of-the-art results.

Findings

MUSE outperforms 14 methods across 10 datasets.

Reconstruction flip challenges existing assumptions in GLAD.

Multifaceted error summaries enhance anomaly detection features.

Abstract

Graph autoencoders (Graph-AEs) learn representations of given graphs by aiming to accurately reconstruct them. A notable application of Graph-AEs is graph-level anomaly detection (GLAD), whose objective is to identify graphs with anomalous topological structures and/or node features compared to the majority of the graph population. Graph-AEs for GLAD regard a graph with a high mean reconstruction error (i.e. mean of errors from all node pairs and/or nodes) as anomalies. Namely, the methods rest on the assumption that they would better reconstruct graphs with similar characteristics to the majority. We, however, report non-trivial counter-examples, a phenomenon we call reconstruction flip, and highlight the limitations of the existing Graph-AE-based GLAD methods. Specifically, we empirically and theoretically investigate when this assumption holds and when it fails. Through our analyses, we further argue that, while the reconstruction errors for a given graph are effective features for GLAD, leveraging the multifaceted summaries of the reconstruction errors, beyond just mean, can further strengthen the features. Thus, we propose a novel and simple GLAD method, named MUSE. The key innovation of MUSE involves taking multifaceted summaries of reconstruction errors as graph features for GLAD. This surprisingly simple method obtains SOTA performance in GLAD, performing best overall among 14 methods across 10 datasets.