Transfer of Learned Heuristics among Planners∗

This paper presents a study on the transfer of learned control knowledge between two different planning techniques. We automatically learn heuristics (usually, in planning, heuristics are also named control knowledge) from one planner search process and apply them to a different planner. The goal is to improve this second planner efficiency solving new problems, i.e. to reduce computer resources (time and memory) during the search, or to improve quality solutions. The learning component is based on a deductive learning method (EBL) that is able to automatically acquire control knowledge by generating bounded explanations of the problem solving episodes in a graph-plan based planner. Then, we transform the learned knowledge so that it can be used by a bidirectional planner.