NeurIPS'19: Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting (Official Pytorch implementation for noisy labels). The implementation of class imbalance is available at https://github.com/xjtushujun/Meta-weight-net_class-imbalance.
================================================================================================================================================================
This is the code for the paper:
Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting
Jun Shu, Qi Xie, Lixuan Yi, Qian Zhao, Sanping Zhou, Zongben Xu, Deyu Meng*
To be presented at NeurIPS 2019.
If you find this code useful in your research then please cite
@inproceedings{han2018coteaching,
title={Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting},
author={Shu, Jun and Xie, Qi and Yi, Lixuan and Zhao, Qian and Zhou, Sanping and Xu, Zongben and Meng, Deyu},
booktitle={NeurIPS},
year={2019}
}
The requiring environment is as bellow:
- Linux
- Python 3+
- PyTorch 0.4.0
- Torchvision 0.2.0
Here is an example:
python train_WRN-28-10_Meta_PGC.py --dataset cifar10 --corruption_type unif(flip2) --corruption_prob 0.6
The default network structure is WRN-28-10, if you want to train with ResNet32 model, please reset the learning rate delay policy.
A stable version is relased.
python MW-Net.py --dataset cifar10 --corruption_type unif(flip2) --corruption_prob 0.6
The new code on github (https://github.com/ShiYunyi/Meta-Weight-Net_Code-Optimization) has implemented the MW-Net based on the newest pytorch and torchvision version. It rewrites an optimizer to assign non leaf node tensors to model parameters. Thus it does not need to rewrite the nn.Module as this version does. Very thanks for Shi Yunyi ([email protected])!
We thank the Pytorch implementation on glc(https://github.com/mmazeika/glc) and learning-to-reweight-examples(https://github.com/danieltan07/learning-to-reweight-examples).
Contact: Jun Shu ([email protected]); Deyu Meng([email protected]).