Confidence-aware Contrastive Learning for Selective Classification

This is the official implementation of the paper Confidence-aware Contrastive Learning for Selective Classification.

Based on our theoretical analysis, in this work, we propose a novel confidence-aware contrastive learning method for selective classification that explicitly improve the selective classification model at the feature level.

Requirements

matplotlib==3.5.3
numpy==1.21.6
Pillow==9.4.0
progress==1.6
scipy==1.10.0
torch==1.13.1
torchvision==0.14.1
tqdm==4.64.1

Training and Evaluation

Run CCL-SC:

bash run_${dataset}_csc.sh

Citation

@inproceedings{
    CCL-SC,
    title={Confidence-aware Contrastive Learning for Selective Classification},
    author={Yu-Chang Wu, Shen-Huan Lyu, Haopu Shang, Xiangyu Wang, Chao Qian},
    booktitle={International Conference on Machine Learning},
    year={2024}
}

Acknowledgement

This code is based on the official code base of SAT+EM (which is based on the official code base of Deep Gambler and Self-Adaptive Training).