LiJunnan1992 / DivideMix

Code for paper: DivideMix: Learning with Noisy Labels as Semi-supervised Learning
MIT License
543 stars 84 forks source link

DivideMix without augmentation #31

Closed QMiao-cs closed 3 years ago

QMiao-cs commented 3 years ago

First, thank you for sharing your code. Could you please explain the details of the diviemix without augmentation in ablation study? Are unlabeled samples also without the data augmentation ?