Hierarchical Relative Entropy Policy Search

Many real-world problems are inherently hierarchically structured. The use of this structure in an agent’s policy may well be the key to improved scalability and higher performance. However, such hierarchical structures cannot be exploited by current policy search algorithms. We will concentrate on a basic, but highly relevant hierarchy — the ‘mixed option’ policy. Here, a gating network first decides which of the options to execute and, subsequently, the option-policy determines the action. In this paper, we reformulate learning a hierarchical policy as a latent variable estimation problem and subsequently extend the Relative Entropy Policy Search (REPS) to the latent variable case. We show that our Hierarchical REPS can learn versatile solutions while also showing an increased performance in terms of learning speed and quality of the found policy in comparison to the nonhierarchical approach.

• Type: Journal
• Year: 2012
• Where: JMLR
• Authors: Christian Daniel, Gerhard Neumann, Jan Peters
• Keywords: Estimation Problem, Gating Network, JMLR 2012, Programming Languages, Relative Entropy