Heuristic Refinements of Approximate Linear Programming for Factored Continuous-State Markov Decision Processes

Approximate linear programming (ALP) offers a promising framework for solving large factored Markov decision processes (MDPs) with both discrete and continuous states. Successful application of the approach depends on the choice of an appropriate set of feature functions defining the value function, and efficient methods for generating constraints that determine the convex space of the solution. The application of the ALP in continuous state-space settings poses an additional challenge

• Type: Conference
• Year: 2004
• Where: AIPS
• Authors: Branislav Kveton, Milos Hauskrecht
• Keywords: AIPS 2004, Appropriate Set, Artificial Intelligence, Continuous States, Large Factored Markov