Online Episodic Convex Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Moreno, Bianca Marin, Eldowa, Khaled, Gaillard, Pierre, Brégère, Margaux, Oudjane, Nadia
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912370646843392
author Moreno, Bianca Marin
Eldowa, Khaled
Gaillard, Pierre
Brégère, Margaux
Oudjane, Nadia
author_facet Moreno, Bianca Marin
Eldowa, Khaled
Gaillard, Pierre
Brégère, Margaux
Oudjane, Nadia
contents We study online learning in episodic finite-horizon Markov decision processes (MDPs) with convex objective functions, known as the concave utility reinforcement learning (CURL) problem. This setting generalizes RL from linear to convex losses on the state-action distribution induced by the agent's policy. The non-linearity of CURL invalidates classical Bellman equations and requires new algorithmic approaches. We introduce the first algorithm achieving near-optimal regret bounds for online CURL without any prior knowledge on the transition function. To achieve this, we use an online mirror descent algorithm with varying constraint sets and a carefully designed exploration bonus. We then address for the first time a bandit version of CURL, where the only feedback is the value of the objective function on the state-action distribution induced by the agent's policy. We achieve a sub-linear regret bound for this more challenging problem by adapting techniques from bandit convex optimization to the MDP setting.
format Preprint
id arxiv_https___arxiv_org_abs_2505_07303
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Online Episodic Convex Reinforcement Learning
Moreno, Bianca Marin
Eldowa, Khaled
Gaillard, Pierre
Brégère, Margaux
Oudjane, Nadia
Machine Learning
We study online learning in episodic finite-horizon Markov decision processes (MDPs) with convex objective functions, known as the concave utility reinforcement learning (CURL) problem. This setting generalizes RL from linear to convex losses on the state-action distribution induced by the agent's policy. The non-linearity of CURL invalidates classical Bellman equations and requires new algorithmic approaches. We introduce the first algorithm achieving near-optimal regret bounds for online CURL without any prior knowledge on the transition function. To achieve this, we use an online mirror descent algorithm with varying constraint sets and a carefully designed exploration bonus. We then address for the first time a bandit version of CURL, where the only feedback is the value of the objective function on the state-action distribution induced by the agent's policy. We achieve a sub-linear regret bound for this more challenging problem by adapting techniques from bandit convex optimization to the MDP setting.
title Online Episodic Convex Reinforcement Learning
topic Machine Learning
url https://arxiv.org/abs/2505.07303