The codes of Robust Causal Graph Representation Learning against Confounding Effects
Please use the python scripts in /RCGRL/spmotif_gen/ to generate the raw data first, then run /RCGRL/train/sp-mtf_rcgrl.py to repeat our experiments. Specificially:
Run python3 /RCGRL/spmotif_gen/spmotif_test_dataset_gen.py to generate test.npy, Run python3 /RCGRL/spmotif_gen/spmotif_train_dataset_gen.py to generate the train.npy, Run python3 /RCGRL/spmotif_gen/spmotif_validate_dataset_gen.py to generate the val.npy,
then put test.npy, train.npy, val.npy into /RCGRL/data/SPMotif-0.9/raw/, and run: python3 /RCGRL/train/sp-mtf_rcgrl.py
If you find this code useful, please cite our paper 'Robust Causal Graph Representation Learning against Confounding Effects'