CONFES : Label Noise-Robust Learning Using a Confidence-Based Sieving Strategy

CONFES is a learning approach that is robust to label noise. It is a novel sieving strategy which detects the samples with label noise and excludes them from the training process.

Installation

First, install the dependencies:

pip3 install -r requirements.txt  -f https://download.pytorch.org/whl/torch_stable.html

Run

Then, you can run CONFES:

python3 main.py --dataset cifar100 --model preact-resnet18 --lr 0.02 --weight-decay 5e-4 --batch-size 128  --epochs 300 --noise-rate 0.5

Citation

If you find the provided code useful for your research, please consider citing our paper:

@article{
torkzadehmahani2023label,
title={Label Noise-Robust Learning using a Confidence-Based Sieving Strategy},
author={Reihaneh Torkzadehmahani and Reza Nasirigerdeh and Daniel Rueckert and Georgios Kaissis},
journal={Transactions on Machine Learning Research},
year={2023},
url={https://openreview.net/forum?id=3taIQG4C7H},
note={}
}