Approximate Value Trees in Structured Dynamic Programming

We propose and examine a method of approximate dynamic programming for Markov decision processes based on structured problem representations. We assume an MDP is represented using a dynamic Bayesian network, and construct value functions using decision trees as our function representation. The size of the representation is kept within acceptable limits by pruning these value trees so that leaves represent possible ranges of values, thus approximating the value functions produced during optimization. We propose a method for detecting convergence,prove errors bounds on the resulting approximately optimal value functions and policies, and describe some preliminary experimental results.