Reinforcement Learning with Stepwise Fairness Constraints

Zhun Deng, He Sun, Steven Wu, Linjun Zhang, David Parkes
Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, PMLR 206:10594-10618, 2023.

Abstract

AI methods are used in societally important settings, ranging from credit to employment to housing, and it is crucial to provide fairness in regard to automated decision making. Moreover, many settings are dynamic, with populations responding to sequential decision policies. We introduce the study of reinforcement learning (RL) with stepwise fairness constraints, which require group fairness at each time step. In the case of tabular episodic RL, we provide learning algorithms with strong theoretical guarantees in regard to policy optimality and fairness violations. Our framework provides tools to study the impact of fairness constraints in sequential settings and brings up new challenges in RL.

Cite this Paper

BibTeX
@InProceedings{pmlr-v206-deng23a, title = {Reinforcement Learning with Stepwise Fairness Constraints}, author = {Deng, Zhun and Sun, He and Wu, Steven and Zhang, Linjun and Parkes, David}, booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics}, pages = {10594--10618}, year = {2023}, editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem}, volume = {206}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v206/deng23a/deng23a.pdf}, url = {https://proceedings.mlr.press/v206/deng23a.html}, abstract = {AI methods are used in societally important settings, ranging from credit to employment to housing, and it is crucial to provide fairness in regard to automated decision making. Moreover, many settings are dynamic, with populations responding to sequential decision policies. We introduce the study of reinforcement learning (RL) with stepwise fairness constraints, which require group fairness at each time step. In the case of tabular episodic RL, we provide learning algorithms with strong theoretical guarantees in regard to policy optimality and fairness violations. Our framework provides tools to study the impact of fairness constraints in sequential settings and brings up new challenges in RL.} }
Endnote
%0 Conference Paper %T Reinforcement Learning with Stepwise Fairness Constraints %A Zhun Deng %A He Sun %A Steven Wu %A Linjun Zhang %A David Parkes %B Proceedings of The 26th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2023 %E Francisco Ruiz %E Jennifer Dy %E Jan-Willem van de Meent %F pmlr-v206-deng23a %I PMLR %P 10594--10618 %U https://proceedings.mlr.press/v206/deng23a.html %V 206 %X AI methods are used in societally important settings, ranging from credit to employment to housing, and it is crucial to provide fairness in regard to automated decision making. Moreover, many settings are dynamic, with populations responding to sequential decision policies. We introduce the study of reinforcement learning (RL) with stepwise fairness constraints, which require group fairness at each time step. In the case of tabular episodic RL, we provide learning algorithms with strong theoretical guarantees in regard to policy optimality and fairness violations. Our framework provides tools to study the impact of fairness constraints in sequential settings and brings up new challenges in RL.
APA
Deng, Z., Sun, H., Wu, S., Zhang, L. & Parkes, D.. (2023). Reinforcement Learning with Stepwise Fairness Constraints. Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 206:10594-10618 Available from https://proceedings.mlr.press/v206/deng23a.html.

Related Material