2026-06-07-millard26a.md

title	Can Optimal Transport Improve Federated Inverse Reinforcement Learning?
section	Poster
openreview	Al2BLZmPX9
abstract	In robotics and multi-agent systems, fleets of autonomous agents often operate in subtly different environments while pursuing a common high-level objective. Directly pooling their data to learn a shared reward function is typically impractical due to differences in dynamics, privacy constraints, and limited communication bandwidth. This paper introduces an optimal transport–based approach to federated inverse reinforcement learning (IRL). Each client first performs lightweight Maximum Entropy IRL locally, adhering to its computational and privacy limitations. The resulting reward functions are then fused via a Wasserstein barycenter, which considers their underlying geometric structure. We further prove that this barycentric fusion yields a more faithful global reward estimate than conventional parameter averaging methods in federated learning. Overall, this work provides a principled and communication-efficient framework for deriving a shared reward that generalizes across heterogeneous agents and environments.
layout	inproceedings
series	Proceedings of Machine Learning Research
publisher	PMLR
issn	2640-3498
id	millard26a
month	0
tex_title	Can Optimal Transport Improve Federated Inverse Reinforcement Learning?
firstpage	1939
lastpage	1953
page	1939-1953
order	1939
cycles	false
bibtex_author	Millard, David and Baheri, Ali
author	given family David Millard given family Ali Baheri
date	2026-06-07
address
container-title	Proceedings of The 8th Annual Learning for Dynamics and Control Conference
volume	331
genre	inproceedings
issued	date-parts 2026 6 7
pdf	https://raw.githubusercontent.com/mlresearch/v331/main/assets/millard26a/millard26a.pdf
extras