Iterative Option Discovery for Planning, by Planning
This addresses the challenge of scaling reinforcement learning and planning to complex domains by improving option discovery, though it is incremental as it builds on the Expert Iteration approach from AlphaZero.
The paper tackles the problem of discovering useful temporal abstractions (options) for reinforcement learning and planning by proposing Option Iteration, which learns a set of locally strong option policies instead of a single global policy. The result is a significant benefit in challenging planning environments compared to using primitive actions and a single rollout policy, as demonstrated experimentally.
Discovering useful temporal abstractions, in the form of options, is widely thought to be key to applying reinforcement learning and planning to increasingly complex domains. Building on the empirical success of the Expert Iteration approach to policy learning used in AlphaZero, we propose Option Iteration, an analogous approach to option discovery. Rather than learning a single strong policy that is trained to match the search results everywhere, Option Iteration learns a set of option policies trained such that for each state encountered, at least one policy in the set matches the search results for some horizon into the future. Intuitively, this may be significantly easier as it allows the algorithm to hedge its bets compared to learning a single globally strong policy, which may have complex dependencies on the details of the current state. Having learned such a set of locally strong policies, we can use them to guide the search algorithm resulting in a virtuous cycle where better options lead to better search results which allows for training of better options. We demonstrate experimentally that planning using options learned with Option Iteration leads to a significant benefit in challenging planning environments compared to an analogous planning algorithm operating in the space of primitive actions and learning a single rollout policy with Expert Iteration.