Conference paper · 2026

Multivariate Distributional Reinforcement Learning Using Sliced Divergences

Baptiste Debes and Tinne Tuytelaars

Proceedings of the 43rd International Conference on Machine Learning (ICML), Seoul, South Korea

arXiv Poster

Plain-English Summary

A multivariate distributional RL framework that uses sliced divergences to make high-dimensional return distributions easier to compare and learn.

Abstract

Introduces sliced distributional reinforcement learning for multivariate return distributions, using sliced divergences to compare projected return distributions tractably.

BibTeX

@inproceedings{debes2026multivariatedistributionalrl,
  title = {Multivariate Distributional Reinforcement Learning Using Sliced Divergences},
  author = {Debes, Baptiste and Tuytelaars, Tinne},
  booktitle = {International Conference on Machine Learning},
  year = {2026},
  url = {https://arxiv.org/abs/2605.31222},
  doi = {10.48550/arXiv.2605.31222}
}