Approximate Policy Iteration with a Policy Language Bias

We study an approach to policy selection for large relational Markov Decision Processes (MDPs). We consider a variant of approximate policy iteration (API) that replaces the usual value-function learning step with a learning step in policy space. This is advantageous in domains where good policies are easier to represent and learn than the corresponding value functions, which is often the case for the relational MDPs we are interested in. In order to apply API to such problems, we introduce a relational policy language and corresponding learner. In addition, we introduce a new bootstrapping routine for goalbased planning domains, based on random walks. Such bootstrapping is necessary for many large relational MDPs, where reward is extremely sparse, as API is ineffective in such domains when initialized with an uninformed policy. Our experiments show that the resulting system is able to find good policies for a number of classical planning domains and their stochastic variants by sol...

• Type: Conference
• Year: 2003
• Where: NIPS
• Authors: Alan Fern, Sung Wook Yoon, Robert Givan
• Keywords: Large Relational Markov, Large Relational Mdps, NIPS 2003, NIPS 2007, Relational Mdps