2024_ICML_TSS

Offical implementation for Mitigating label noise on graph via topological sample selection (ICML 2024) (PyTorch implementation).

This is the code for the paper: Mitigating Label Noise on Graph via Topological Sample Selection
Yuhao Wu, Jiangchao Yao, Xiaobo Xia, Jun Yu, Ruxin Wang, Bo Han, Tongliang Liu.

If you find this code useful in your research, please cite

@inproceedings{wumitigating,
  title={Mitigating Label Noise on Graphs via Topological Sample Selection},
  author={Wu, Yuhao and Yao, Jiangchao and Xia, Xiaobo and Yu, Jun and Wang, Ruxin and Han, Bo and Liu, Tongliang},
  booktitle={Forty-first International Conference on Machine Learning}
}

Dependencies

we implement our methods by PyTorch on NVIDIA RTX 4090. The environment is as bellow:

• PyTorch, version == 2.2.2
• CUDA, version == 11.8

Experiments

Here is an example:

python main.py --dataset cora --noise_rate 0.5