Elucidating Robust Learning with Uncertainty-Aware Corruption Pattern Estimation
Introduction
Elucidating Robust Learning with Uncertainty-Aware Corruption Pattern Estimation
Our contributions are as follows
Objective
Architecture
Requirements
torch==1.7.1
torchvision==0.8.2
matplotlib==3.4.1
scikit-learn==0.24.1
gensim==4.0.1
scipy==1.6.2
seaborn==0.11.1
Pillow==8.2.0
Datasets
Please download mannually TREC dataset
TREC TREC
Reproducing results of the paper
e.g., mnist on class conditional noise setting
mkdir ckpt
mkdir res
cd scripts
./ccn_mnist.sh
๐ก
Class Conditional Noise
CIFAR10
Flipping Rate | F-correction | Co-teaching | Co-teaching+ | JoCoR | MLN(ours) |
---|---|---|---|---|---|
Symmetry-20% | 68.74ยฑ0.20 | 78.23ยฑ0.27 | 78.71ยฑ0.34 | 85.73ยฑ0.19 | 84.20ยฑ0.05 |
Symmetry-50% | 42.71ยฑ0.42 | 71.30ยฑ0.13 | 57.05ยฑ0.54 | 79.41ยฑ0.25 | 77.88ยฑ0.07 |
Symmetry-80% | 15.88ยฑ0.42 | 26.58ยฑ2.22 | 24.19ยฑ2.74 | 27.78ยฑ3.06 | 41.83ยฑ0.10 |
Asymmetry-40% | 70.60ยฑ0.40 | 73.78ยฑ0.22 | 68.84ยฑ0.20 | 76.36ยฑ0.49 | 76.62ยฑ0.07 |
Noise Transition Matrix on CIFAR10
๐ก
Set Dependent Noise
aleatoric uncertainty for the ambiguous set is higher than the clean set and larger for more label noise rate.
estimated noise transition matrix for partioned sets are:
Citing our paper
If you find this work useful please consider citing it:
@article{papername,
title={title},
author={authors},
journal={arXiv preprint arXiv:xxxx.xxxxx},
year={2021}
}