Efficient Reinforcement Learning in Probabilistic Reward Machines
Abstract
In this paper, we study reinforcement learning in Markov Decision Processes with Probabilistic Reward Machines (PRMs), a form of non-Markovian reward commonly found in robotics tasks. We design an algorithm for PRMs that achieves a regret bound of O(HOAT + H2O2A3/2 + HT), where H is the time horizon, O is the number of observations, A is the number of actions, and T is the number of time-steps. This result improves over the best-known bound, O(HOAT) of pmlr-v206-bourel23a for MDPs with Deterministic Reward Machines (DRMs), a special case of PRMs. When T ≥ H3O3A2 and OA ≥ H, our regret bound leads to a regret of O(HOAT), which matches the established lower bound of (HOAT) for MDPs with DRMs up to a logarithmic factor. To the best of our knowledge, this is the first efficient algorithm for PRMs. Additionally, we present a new simulation lemma for non-Markovian rewards, which enables reward-free exploration for any non-Markovian reward given access to an approximate planner. Complementing our theoretical findings, we show through extensive experiment evaluations that our algorithm indeed outperforms prior methods in various PRM environments.
Turn this paper into a lesson
ArcXiv compiles a structured reading guide from this paper's metadata: plain-English importance, contributions, prerequisite concepts, which sections to read first, flashcards, and a quiz. Grounded in the abstract, never invented.