Skip to content

mndu/RNF-Fairness

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Fairness via Representation Neutralization

PyTorch code for the Neurips 2021 paper: Fairness via Representation Neutralization. In this code, we use MEPS dataset as example.

Usage Instructions:

To run RNF (with proxy attribute annotations):

python train_rnf.py 

The hyperparameter alpha in the train_rnf.py file is used to control the fairness accuracy trade-off. For MEPS dataset, a reasonable range for the alpha value is between [0, 0.035].

System requirement:

torch==0.4.1.post2, torchtext==0.2.3

Reference:

@inproceedings{du2021fairness,
  title={Fairness via Representation Neutralization},
  author={Du, Mengnan and Mukherjee, Subhabrata and Wang, Guanchu and Tang, Ruixiang and Awadallah, Ahmed Hassan and Hu, Xia},
  booktitle={Neurips},
  year={2021}
}

About

PyTorch code for the Neurips 2021 paper: Fairness via Representation Neutralization

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages