A New View on Planning in Online Reinforcement Learning
This addresses a computational bottleneck in reinforcement learning for AI systems, though it is an incremental improvement over existing methods.
The paper tackles the inefficiency of background planning in model-based reinforcement learning by introducing goal-space planning (GSP), which constrains planning to subgoals and uses local models, resulting in faster learning for base learners across domains.
This paper investigates a new approach to model-based reinforcement learning using background planning: mixing (approximate) dynamic programming updates and model-free updates, similar to the Dyna architecture. Background planning with learned models is often worse than model-free alternatives, such as Double DQN, even though the former uses significantly more memory and computation. The fundamental problem is that learned models can be inaccurate and often generate invalid states, especially when iterated many steps. In this paper, we avoid this limitation by constraining background planning to a set of (abstract) subgoals and learning only local, subgoal-conditioned models. This goal-space planning (GSP) approach is more computationally efficient, naturally incorporates temporal abstraction for faster long-horizon planning and avoids learning the transition dynamics entirely. We show that our GSP algorithm can propagate value from an abstract space in a manner that helps a variety of base learners learn significantly faster in different domains.