Towards Effective and General Graph Unlearning via Mutual Evolution

Requirements

Hardware environment: Intel(R) Xeon(R) CPU E5-2686 v4 @ 2.30GHz, NVIDIA GeForce RTX 3090 with 24GB.

Software environment: Ubuntu 20.04.5, Python 3.8.10, Pytorch 1.13.0, and CUDA 11.7.0.

1. Please refer to PyTorch and PyG to install the environments;

2. Run 'pip install -r requirements.txt' to download required packages;

Training To train model(s) in the paper

1. Please unzip xxx.zip to the current file directory location

2. Please refer to the configs folds to modify the hyperparameters

   config.py - Setting the path for data loading and saving.

   parameter_parser.py - Parameter settings for modeling and training.

3. Open main.py to start unlearning

   We provide Cora dataset as example.

   Meanwhile, you can personalize your settings (lib_dataset/lib_gnn_model/exp).

   Run this command:

python main.py