FraudGT: A Simple, Effective, and Efficient Graph Transformer for Financial Fraud Detection

This repository holds the code for FraudGT framework.

Environment Setup

You can create a conda environment to easily run the code. For example, we can create a virtual environment named fraudGT:

conda create -n fraudGT python=3.9 -y
conda activate fraudGT

Install the required packages using the following commands:

conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia
conda install pyg -c pyg
pip install -r requirements.txt

Run the Code

You will need to firstly specify the dataset path (./data in this example) and log location (./results in this example) by editing the config file provided under ./configs/{dataset_name}/. An example configuration is

......
out_dir: ./results
dataset:
  dir: ./data
......

Download and unzip the Anti-Money Laundering dataset into your specified dataset path (for example, put the unzipped HI-small.csv into ./data). Dataset will be automatically processed at the first run.

For convenience, a script file is created to run the experiment with specified configuration. For instance, you can edit and run the interactive_run.sh to start the experiment.

cd FraudGT
chmox +x ./run/interactive_run.sh
./run/interactive_run.sh