English

Optimistic Curiosity Exploration and Conservative Exploitation with Linear Reward Shaping

Machine Learning 2022-10-18 v2 Artificial Intelligence

Abstract

In this work, we study the simple yet universally applicable case of reward shaping in value-based Deep Reinforcement Learning (DRL). We show that reward shifting in the form of the linear transformation is equivalent to changing the initialization of the QQ-function in function approximation. Based on such an equivalence, we bring the key insight that a positive reward shifting leads to conservative exploitation, while a negative reward shifting leads to curiosity-driven exploration. Accordingly, conservative exploitation improves offline RL value estimation, and optimistic value estimation improves exploration for online RL. We validate our insight on a range of RL tasks and show its improvement over baselines: (1) In offline RL, the conservative exploitation leads to improved performance based on off-the-shelf algorithms; (2) In online continuous control, multiple value functions with different shifting constants can be used to tackle the exploration-exploitation dilemma for better sample efficiency; (3) In discrete control tasks, a negative reward shifting yields an improvement over the curiosity-based exploration method.

Keywords

Cite

@article{arxiv.2209.07288,
  title  = {Optimistic Curiosity Exploration and Conservative Exploitation with Linear Reward Shaping},
  author = {Hao Sun and Lei Han and Rui Yang and Xiaoteng Ma and Jian Guo and Bolei Zhou},
  journal= {arXiv preprint arXiv:2209.07288},
  year   = {2022}
}
R2 v1 2026-06-28T01:21:50.659Z