DPSBA

The source code for "Stealthy Yet Effective: Distribution-Preserving Backdoor Attacks on Graph Classification" at NeurIPS 2025.

Part of code is referenced from paper Graph Backdoor.

Datasets

You can download the datasets from TUDataset

Requirements

• Python >= 3.9
• dgl = 2.2.1
• numpy = 1.26.4
• torch = 2.1.2+cu121
• torch-geometric = 2.6.1
• tqdm = 4.64.1

Run

cd main
python attackk22.py --use_org_node_attr --save_clean_model --save_bkd_model

Cite

Please cite our paper if it is helpful in your own work: Stealthy Yet Effective: Distribution-Preserving Backdoor Attacks on Graph Classification