Exploration--Exploitation in MDPs with Options
arXiv:1703.08667
Abstract
While a large body of empirical results show that temporally-extended actions and options may significantly affect the learning performance of an agent, the theoretical understanding of how and when options can be beneficial in online reinforcement learning is relatively limited. In this paper, we derive an upper and lower bound on the regret of a variant of UCRL using options. While we first analyze the algorithm in the general case of semi-Markov decision processes (SMDPs), we show how these results can be translated to the specific case of MDPs with options and we illustrate simple scenarios in which the regret of learning with options can be \textit{provably} much smaller than the regret suffered when learning with primitive actions.
Cited by in corpus (6)
- Approximate Exploration through State Abstraction
- Provable Hierarchical Imitation Learning via EM
- Hierarchical Cross-Modal Agent for Robotics Vision-and-Language Navigation
- Temporal Abstraction in Reinforcement Learning with the Successor Representation
- Average-Reward Learning and Planning with Options
- Provable Hierarchy-Based Meta-Reinforcement Learning