Skip to content
/ FATDM Public

Fairness and Accuracy Transfer by Density Matching

Notifications You must be signed in to change notification settings

pth1993/FATDM

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

FATDM - Fairness and Accuracy Transfer by Density Matching


Code by Thai-Hoang Pham at Ohio State University.

1. Introduction

This repository contains source code (FATDM) and data for paper "Fairness and Accuracy under Domain Generalization" (ICLR 2023)

FATDM is a Pytorch implementation of the two-stage network (see Figure 1) which achieves fair and accurate predictions in unseen target domain (domain generalization) via invariant representation learnings.

alt text

Figure 1: Two-stage learning

2. Installation

FATDM depends on pytorch (CUDA toolkit if use GPU), torchvision, numpy, scipy, tqdm, pandas, scikit-learn. You must have them installed before using FATDM. The simple way to install them is using conda.

    # Using GPU
    $ conda install pytorch torchvision pytorch-cuda=11.7 -c pytorch -c nvidia
    $ conda install numpy scipy tqdm pandas scikit-learn
    # Using CPU
    $ conda install pytorch torchvision cpuonly -c pytorch
    $ conda install numpy scipy tqdm pandas scikit-learn

3. Usage

3.1. Data

The datasets used to train and evaluate FATDM is processed from MIMIC-CXR-JPG (chest radiographs with structured labels) dataset retrieved from PhysioNet. MIMIC-CXR-JPG dataset is restricted-access resource. To access this dataset, user must sign the data use agreement in the project website link.

3.2. Training FATDM

The training script for FATDM are as follows:

  • main_stargan.py: Training script for StarGAN to learn density mapping functions.
  • main_cyclegan.py: Training script for CycleGAN to learn density mapping functions.
  • main_fatdm.py: Training script for FATDM to transfer fairness and accuracy to new domains.

4. References

@inproceedings{
pham2023fairness,
title={Fairness and Accuracy under Domain Generalization},
author={Thai-Hoang Pham and Xueru Zhang and Ping Zhang},
booktitle={The Eleventh International Conference on Learning Representations },
year={2023},
url={https://openreview.net/forum?id=jBEXnEMdNOL}
}

5. Contact

Thai-Hoang Pham < [email protected] >

Department of Computer Science and Engineering, Ohio State University, USA

About

Fairness and Accuracy Transfer by Density Matching

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages