Policy Optimization with Constraint Advantage Regularization (POCAR)

• Policy Optimization with Advantage Regularization for Long-Term Fairness in Decision Systems
  Authors: Eric Yang Yu, Zhizhen Qin, Min Kyung Lee, Sicun Gao
  NeurIPS (Conference on Neural Information Processing Systems) 2022

This is the code implementation for the NeurIPS 2022 paper above. Code and environments are adapted from the original Google ML-fairness-gym repo.

Installation

First, install Anaconda to set up virtual environment.

To use python 3.8, run:

conda create -n pocar python=3.8
conda activate pocar
pip install -r requirements_py38.txt

To use python 3.7, run:

conda create -n pocar python=3.7
conda activate pocar
pip install -r requirements_py37.txt

Usage

First, cd to an experiment directory. Then, to train:

python main.py --train

To view training progress:

python main.py --show_train_progress

To evaluate models (make sure to specify model paths to evaluate in config.py):

python main.py --eval_path ./path/to/evaluations/

To view model evaluations:

python main.py --display_eval_path ./path/to/evaluations/

Environments

Attention Allocation

Lending

Vaccination / Infectious Disease Control

Bibtex

@misc{https://doi.org/10.48550/arxiv.2210.12546,
  doi = {10.48550/ARXIV.2210.12546},
  url = {https://arxiv.org/abs/2210.12546},
  author = {Yu, Eric Yang and Qin, Zhizhen and Lee, Min Kyung and Gao, Sicun},
  keywords = {Machine Learning (cs.LG), Artificial Intelligence (cs.AI), Computers and Society (cs.CY), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Policy Optimization with Advantage Regularization for Long-Term Fairness in Decision Systems},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution 4.0 International}
}