On Learning Action Costs from Input Plans
This addresses a gap in action model learning for planning tasks, though it appears incremental as it builds on existing work on learning action dynamics.
The paper tackles the problem of learning action costs from input plans to rank different plans, introducing a new algorithm LACFIP^k that successfully solves this task with theoretical and empirical validation.
Most of the work on learning action models focus on learning the actions' dynamics from input plans. This allows us to specify the valid plans of a planning task. However, very little work focuses on learning action costs, which in turn allows us to rank the different plans. In this paper we introduce a new problem: that of learning the costs of a set of actions such that a set of input plans are optimal under the resulting planning model. To solve this problem we present $LACFIP^k$, an algorithm to learn action's costs from unlabeled input plans. We provide theoretical and empirical results showing how $LACFIP^k$ can successfully solve this task.