Exploration--Exploitation in MDPs with Options
This provides theoretical insights into the benefits of options in MDPs, addressing a gap in understanding for reinforcement learning researchers, though it appears incremental as it builds on existing UCRL methods.
The paper tackles the problem of understanding when and how options (temporally-extended actions) can improve online reinforcement learning by deriving upper and lower bounds on regret for a variant of UCRL with options, showing that in simple scenarios, regret with options can be provably much smaller than with primitive actions.
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.