Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GLADMamba: Unsupervised Graph-Level Anomaly Detection Powered by Selective State Space Model

Yali Fu, Jindong Li, Qi Wang, Qianli Xing

TL;DR

GLADMamba tackles unsupervised graph-level anomaly detection by integrating selective state space models (Mamba) with explicit spectral information. It introduces View-Fused Mamba (VFM) for multi-view fusion and Spectrum-Guided Mamba (SGM) guided by the Rayleigh quotient to refine embeddings, enabling dynamic focus on anomaly-related patterns. The method demonstrates state-of-the-art performance on 12 real-world datasets, outperforming both traditional kernel-based pipelines and end-to-end GNN/Transformer approaches, with improved efficiency due to linear-time updates. The work pioneers applying Mamba and spectral cues in UGLAD, offering a scalable framework for robust anomaly detection in graphs and potential for broad applicability.

Abstract

Unsupervised graph-level anomaly detection (UGLAD) is a critical and challenging task across various domains, such as social network analysis, anti-cancer drug discovery, and toxic molecule identification. However, existing methods often struggle to capture the long-range dependencies efficiently and neglect the spectral information. Recently, selective State Space Models (SSMs), particularly Mamba, have demonstrated remarkable advantages in capturing long-range dependencies with linear complexity and a selection mechanism. Motivated by their success across various domains, we propose GLADMamba, a novel framework that adapts the selective state space model into UGLAD field. We design View-Fused Mamba (VFM) with a Mamba-Transformer-style architecture to efficiently fuse information from different views with a selective state mechanism. We also design Spectrum-Guided Mamba (SGM) with a Mamba-Transformer-style architecture to leverage the Rayleigh quotient to guide the embedding refining process. GLADMamba can dynamically focus on anomaly-related information while discarding irrelevant information for anomaly detection. To the best of our knowledge, this is the first work to introduce Mamba and explicit spectral information to UGLAD. Extensive experiments on 12 real-world datasets demonstrate that GLADMamba outperforms existing state-of-the-art methods, achieving superior performance in UGLAD. The code is available at https://github.com/Yali-F/GLADMamba.

GLADMamba: Unsupervised Graph-Level Anomaly Detection Powered by Selective State Space Model

TL;DR

GLADMamba tackles unsupervised graph-level anomaly detection by integrating selective state space models (Mamba) with explicit spectral information. It introduces View-Fused Mamba (VFM) for multi-view fusion and Spectrum-Guided Mamba (SGM) guided by the Rayleigh quotient to refine embeddings, enabling dynamic focus on anomaly-related patterns. The method demonstrates state-of-the-art performance on 12 real-world datasets, outperforming both traditional kernel-based pipelines and end-to-end GNN/Transformer approaches, with improved efficiency due to linear-time updates. The work pioneers applying Mamba and spectral cues in UGLAD, offering a scalable framework for robust anomaly detection in graphs and potential for broad applicability.

Abstract

Unsupervised graph-level anomaly detection (UGLAD) is a critical and challenging task across various domains, such as social network analysis, anti-cancer drug discovery, and toxic molecule identification. However, existing methods often struggle to capture the long-range dependencies efficiently and neglect the spectral information. Recently, selective State Space Models (SSMs), particularly Mamba, have demonstrated remarkable advantages in capturing long-range dependencies with linear complexity and a selection mechanism. Motivated by their success across various domains, we propose GLADMamba, a novel framework that adapts the selective state space model into UGLAD field. We design View-Fused Mamba (VFM) with a Mamba-Transformer-style architecture to efficiently fuse information from different views with a selective state mechanism. We also design Spectrum-Guided Mamba (SGM) with a Mamba-Transformer-style architecture to leverage the Rayleigh quotient to guide the embedding refining process. GLADMamba can dynamically focus on anomaly-related information while discarding irrelevant information for anomaly detection. To the best of our knowledge, this is the first work to introduce Mamba and explicit spectral information to UGLAD. Extensive experiments on 12 real-world datasets demonstrate that GLADMamba outperforms existing state-of-the-art methods, achieving superior performance in UGLAD. The code is available at https://github.com/Yali-F/GLADMamba.

Paper Structure

This paper contains 23 sections, 24 equations, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Graph-Level Anomaly Detection
  4. State Space Models
  5. Preliminaries
  6. Problem Statement
  7. State Space Models & Mamba
  8. Rayleigh Quotient
  9. Methodology
  10. Data Augmentation and Encoding
  11. View-Fused Mamba (VFM)
  12. Spectrum-Guided Mamba (SGM)
  13. Training and Inference
  14. Experiments
  15. Experiment Settings
  16. ...and 8 more sections

Figures (5)

  • Figure 1: The key factors related to UGLAD.
  • Figure 2: Overview of the proposed GLADMamba model: (a) illustrates the overall pipeline of the framework; (b) depicts the View-Fused Mamba (VFM) module, which integrates multi-view information; (c) shows the Spectrum-Guided Mamba (SGM) module, which is designed to guide embedding refining process by Rayleigh quotient.
  • Figure 3: Ablation study on key components on representative datasets.
  • Figure 4: The hyper-parameter analysis on representative datasets.
  • Figure 5: Visualization analysis on AIDS dataset. ($\bullet$ normal graph, $\bullet$ anomalous graph.)