dqwang122 / GCFExplainer

Global Counterfactual Explainer for Graph Neural Networks

Global Counterfactual Explainer for Graph Neural Networks

This repository is a reference implementation of the global graph counterfactual explainer as described in the paper:

Global Counterfactual Explainer for Graph Neural Networks.
Mert Kosan*, Zexi Huang*, Sourav Medya, Sayan Ranu, Ambuj Singh.
ACM International Conference on Web Search and Data Mining, 2023.

• All codes and datasets are tested with Python 3.8.0, PyTorch 1.7.1, PyTorchGeometric 1.7.0, NumPY 1.21.4, NetworkX 2.5, tqdm 4.53.0.
• We run our experiments on a machine with 2 NVIDIA GeForce RTX 2080 GPU (8GB of RAM) and 32 Intel Xeon CPUs (2.10GHz and 128GB of RAM).

Important files

• gnn.py: training gnn models, we already shared trained ones at data/{dataset_name}/gnn
• vrrw.py: To generate counterfactuals
• summary.py: To generate summary from counterfactuals

Sample command to run dataset for counterfactual and summary generation

python vrrw.py --dataset <dataset_name>

python summary.py --dataset <dataset_name>