English

Utilizing Prior Solutions for Reward Shaping and Composition in Entropy-Regularized Reinforcement Learning

Machine Learning 2023-01-02 v2

Abstract

In reinforcement learning (RL), the ability to utilize prior knowledge from previously solved tasks can allow agents to quickly solve new problems. In some cases, these new problems may be approximately solved by composing the solutions of previously solved primitive tasks (task composition). Otherwise, prior knowledge can be used to adjust the reward function for a new problem, in a way that leaves the optimal policy unchanged but enables quicker learning (reward shaping). In this work, we develop a general framework for reward shaping and task composition in entropy-regularized RL. To do so, we derive an exact relation connecting the optimal soft value functions for two entropy-regularized RL problems with different reward functions and dynamics. We show how the derived relation leads to a general result for reward shaping in entropy-regularized RL. We then generalize this approach to derive an exact relation connecting optimal value functions for the composition of multiple tasks in entropy-regularized RL. We validate these theoretical contributions with experiments showing that reward shaping and task composition lead to faster learning in various settings.

Keywords

Cite

@article{arxiv.2212.01174,
  title  = {Utilizing Prior Solutions for Reward Shaping and Composition in Entropy-Regularized Reinforcement Learning},
  author = {Jacob Adamczyk and Argenis Arriojas and Stas Tiomkin and Rahul V. Kulkarni},
  journal= {arXiv preprint arXiv:2212.01174},
  year   = {2023}
}

Comments

Conference paper accepted in the Main track for AAAI-2023

R2 v1 2026-06-28T07:20:27.599Z