Saved in:
Bibliographic Details
Main Authors: Veviurko, Grigorii, Böhmer, Wendelin, de Weerdt, Mathijs
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.01361
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929727925649408
author Veviurko, Grigorii
Böhmer, Wendelin
de Weerdt, Mathijs
author_facet Veviurko, Grigorii
Böhmer, Wendelin
de Weerdt, Mathijs
contents In reinforcement learning (RL), different reward functions can define the same optimal policy but result in drastically different learning performance. For some, the agent gets stuck with a suboptimal behavior, and for others, it solves the task efficiently. Choosing a good reward function is hence an extremely important yet challenging problem. In this paper, we explore an alternative approach for using rewards for learning. We introduce \textit{max-reward RL}, where an agent optimizes the maximum rather than the cumulative reward. Unlike earlier works, our approach works for deterministic and stochastic environments and can be easily combined with state-of-the-art RL algorithms. In the experiments, we study the performance of max-reward RL algorithms in two goal-reaching environments from Gymnasium-Robotics and demonstrate its benefits over standard RL. The code is available at https://github.com/veviurko/To-the-Max.
format Preprint
id arxiv_https___arxiv_org_abs_2402_01361
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle To the Max: Reinventing Reward in Reinforcement Learning
Veviurko, Grigorii
Böhmer, Wendelin
de Weerdt, Mathijs
Machine Learning
In reinforcement learning (RL), different reward functions can define the same optimal policy but result in drastically different learning performance. For some, the agent gets stuck with a suboptimal behavior, and for others, it solves the task efficiently. Choosing a good reward function is hence an extremely important yet challenging problem. In this paper, we explore an alternative approach for using rewards for learning. We introduce \textit{max-reward RL}, where an agent optimizes the maximum rather than the cumulative reward. Unlike earlier works, our approach works for deterministic and stochastic environments and can be easily combined with state-of-the-art RL algorithms. In the experiments, we study the performance of max-reward RL algorithms in two goal-reaching environments from Gymnasium-Robotics and demonstrate its benefits over standard RL. The code is available at https://github.com/veviurko/To-the-Max.
title To the Max: Reinventing Reward in Reinforcement Learning
topic Machine Learning
url https://arxiv.org/abs/2402.01361