Near-Optimal Representation Learning for Hierarchical Reinforcement Learning
This addresses the challenge of choosing effective goal representations in hierarchical RL, which is crucial for improving performance in complex control tasks, though it appears incremental as it builds on existing hierarchical frameworks.
The paper tackles the problem of representation learning in hierarchical reinforcement learning by developing a notion of sub-optimality and deriving bounds that translate into practical objectives, resulting in qualitatively better representations and quantitatively better hierarchical policies on continuous-control tasks.
We study the problem of representation learning in goal-conditioned hierarchical reinforcement learning. In such hierarchical structures, a higher-level controller solves tasks by iteratively communicating goals which a lower-level policy is trained to reach. Accordingly, the choice of representation -- the mapping of observation space to goal space -- is crucial. To study this problem, we develop a notion of sub-optimality of a representation, defined in terms of expected reward of the optimal hierarchical policy using this representation. We derive expressions which bound the sub-optimality and show how these expressions can be translated to representation learning objectives which may be optimized in practice. Results on a number of difficult continuous-control tasks show that our approach to representation learning yields qualitatively better representations as well as quantitatively better hierarchical policies, compared to existing methods (see videos at https://sites.google.com/view/representation-hrl).