shell
git clone https://github.com/MyCurveNet/CurveNet.git
cd CurveNet
cd torchlight
python setup.py install
cd ..
- In the probing stage, obtain the
python main.py recognition_cycle -c config/cycle/cifar10_unif/train_c0.4_imb0.05.yaml
- In the allocating stage
python main.py recognition_curve -c config/aug/cifar10/train_c0.4_imb0.05.yaml
After training, you can visualize the weight.
python python tools/gen_fig1_weight_cls.py work_dir/recognition_cycle/cifar10_unif/cor0.4_imb0.05 work_dir/recognition_aug_cycle/cifar10_unif/cor0.4_imb0.05
If CurveNet is useful for your research, please consider citing:
@inproceedings{jiang2022delving,
title={Delving into Sample Loss Curve to Embrace Noisy and Imbalanced Data},
author={Jiang, Shenwang and Li, Jianan and Wang, Ying and Huang, Bo and Zhang, Zhang and Xu, Tingfa},
booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
year={2022}
}
- The code is based on Meta-weight-net.
- ResNet-32 is from MLC