Uniform Risk Minimization

Official code for the TMLR paper "Uniformly Distributed Feature Representations for Fair and Robust Learning": https://openreview.net/forum?id=PgLbS5yp8n

Domain Generalization

URM is available as part of the DomainBed library. Alternatively, you may use this fork of the DomainBed library to run URM for domain generalization. Passing --algorithm URM to the commands will run the URM algorithm when running experiments.

Subpopulation robustness

Please use this fork of the SubpopBench library to run URM for subpopulation shifts. Passing --algorithm URM to the commands will run the URM algorithm when running experiments. We have also created a pull request to merge the forked repo to the original SubpopBench library. You may check the original SubpopBench library in case the pull request has been merged.

Citation

If you find this code or idea useful, please cite our work.

@article{
krishnamachari2024uniformly,
title={Uniformly Distributed Feature Representations for Fair and Robust Learning},
author={Kiran Krishnamachari and See-Kiong Ng and Chuan-Sheng Foo},
journal={Transactions on Machine Learning Research},
issn={2835-8856},
year={2024},
url={https://openreview.net/forum?id=PgLbS5yp8n},
note={}
}