OT-Cleaner

Official code of OT-Cleaner presented at ICASSP 2022.

Model Training

python train.py --dataset cifar10 --noise_rate 0.2 --noise_type symmetric

Citation

@inproceedings{xia2022ot,
  title={Ot cleaner: Label correction as optimal transport},
  author={Xia, Jun and Tan, Cheng and Wu, Lirong and Xu, Yongjie and Li, Stan Z},
  booktitle={ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={3953--3957},
  year={2022},
  organization={IEEE}
}