GAD-EBM: Graph Anomaly Detection
using Energy-Based Models

This repository contains the PyTorch implementation of the NeurIPS 2023 New Frontiers in Graph Learning (GLFrontiers) workshop paper "GAD-EBM: Graph Anomaly Detection using Energy-Based Models" by Amit Roy, Juan Shu, Olivier Elshocht, Jeroen Smeets, Ruqi Zhang and Pan Li.

Abstract

Graph Anomaly Detection (GAD) is essential in fields ranging from network security, and bioinformatics to finance. Previous works often adopt auto-encoders to compute reconstruction errors for anomaly detection: anomalies are hard to be reconstructed. In this work, we revisit the first principle for anomaly detection, i.e., the Neyman-Pearson rule, where the optimal anomaly detector is based on the likelihood of a data point given the normal distribution of data. However, in practice, the distribution is often unknown and the estimation of the distribution of graph-structured data may be hard. Moreover, the likelihood computation of a graph-structured data point may be challenging as well. In this paper, we propose a novel approach GAD-EBM that can estimate the distribution of graphs and compute likelihoods efficiently by using Energy-Based Models (EBMs) over graphs. GAD-EBM approaches the likelihood of a rooted subgraph of node $v$, and further can leverage the likelihood to accurately identify whether node $v$ is anomalous or not. Traditional score matching for training EBMs may not be used to apply EBMs that model the distribution of graphs because of complicated discreteness and multi-modality of graph data. We propose a Subgraph Score Matching (SSM) approach, which is specifically designed for graph data based on a novel framework of neighborhood state-space graphs. Experimentation conducted on six real-world datasets validates the effectiveness and efficiency of GAD-EBM and the source code for GAD-EBM is openly available.

Neighborhood State-Space Graph

Exapmle of a neighborhood state-space graph

Main Parameters

--dataset                    Anomaly detection dataset (default: disney)
--perturb_percent            Percentages of edges to be added/deleted (default: 0.05)
--seed                       Random Number Seed (default: 42)
--nb_epochs                  Number of epochs (default: 200)
--hidden_dim                 Hidden Dimension Size (default: 16)
--lr                         Learning Rate (default: 0.01)
--l2_coef                    Regularization coefficient (default: 0.01)
--drop_edge                  Drop Edge Flag (default: True)
--add_edge                   Add Edge Flag (default: False)
--self_loop                  Self-loop flag (default: True)
--preprocess_feat            Preprocess Features (default: True)
--GNN_name                   GNN Encoder (default: GCN)
--num_neigh                  Number of Neighbors in the State-Space Graph (default: 1)                 

Environment Setup

Create Conda Environment

conda create --name GAD-EBM
conda activate GAD-EBM

Install pytorch:

conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia

Install pytorch geometric:

pip install pyg-lib torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric -f https://data.pyg.org/whl/torch-1.13.0+cu117.html

Install requirements.txt

conda install --file requirements.txt

Basic Usage

Run the python notebook with appropriate parameter changes.

To run GAD-EBM on the DGraph dataset, please download the DGraphFin dataset file 'DGraphFin.zip' from the website 'https://dgraph.xinye.com/introduction' and place it under the directory './dataset/raw'.

Experimental Results

Dataset Description

Benchmark Anomaly Detection Results

Likelihood comparison

Running Time Comparison

Cite

If you find our paper and repo useful, please cite our paper:

@inproceedings{roy2023gad,
  title={GAD-EBM: Graph Anomaly Detection using Energy-Based Models},
  author={Roy, Amit and Shu, Juan and Elshocht, Olivier and Smeets, Jeroen and Zhang, Ruqi and Li, Pan},
  booktitle={NeurIPS 2023 Workshop: New Frontiers in Graph Learning},
  year={2023}
}

Star History