Integrating Machine Learning into Fair Inference

Haoyu Wang, Hanyu Hu, Mingrui Zhuang, Jiayi Shen

2022

Abstract

With the boom of machine learning, fairness is an issue that needs to be concerned. The three main perspectives of this paper provide a thorough look at the fairness problem: First, we introduce a handy tool for causal inference, that is, causal graph, and apply formulas like adjustment formula, back-door formula, and front-door formula to see the effect of interventions, which can help with the fairness. Then some approaches to measure the fairness are introduced: natural direct path and path-specific effect. Finally, we use counterfactual inference further to study fairness with the help of causal graphs and integrate LFR, a model focusing on both group fairness and individual fairness.

Paper Citation

in Harvard Style

Wang H., Hu H., Zhuang M. and Shen J. (2022). Integrating Machine Learning into Fair Inference. In Proceedings of the 2nd International Conference on New Media Development and Modernized Education - Volume 1: NMDME; ISBN 978-989-758-630-9, SciTePress, pages 144-154. DOI: 10.5220/0011908000003613

in Bibtex Style

@conference{nmdme22,
author={Haoyu Wang and Hanyu Hu and Mingrui Zhuang and Jiayi Shen},
title={Integrating Machine Learning into Fair Inference},
booktitle={Proceedings of the 2nd International Conference on New Media Development and Modernized Education - Volume 1: NMDME},
year={2022},
pages={144-154},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011908000003613},
isbn={978-989-758-630-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on New Media Development and Modernized Education - Volume 1: NMDME
TI - Integrating Machine Learning into Fair Inference
SN - 978-989-758-630-9
AU - Wang H.
AU - Hu H.
AU - Zhuang M.
AU - Shen J.
PY - 2022
SP - 144
EP - 154
DO - 10.5220/0011908000003613
PB - SciTePress