arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.10889cs.AI
收藏

从输入规划中学习动作代价

On Learning Action Costs from Input Plans

J.P. Morgan AI Research(摩根大通人工智能研究)

Marianela Morales, Alberto Pozanco, Giuseppe Canonaco, Sriram Gopalakrishnan, Daniel Borrajo, Manuela Veloso

AI总结:

本文提出新问题:学习动作代价使输入规划在所得模型下最优,并设计算法$LACFIP^k$从无标签规划中学习代价,理论和实验验证其有效性。

摘要:

大多数关于学习动作模型的工作集中于从输入规划中学习动作的动态。这使我们能够指定一个规划任务的有效规划。然而,很少有工作关注学习动作代价,而动作代价反过来使我们能够对不同的规划进行排序。在本文中,我们引入一个新问题:学习一组动作的代价,使得一组输入规划在所得的规划模型下是最优的。为了解决这个问题,我们提出了$LACFIP^k$,一种从未标记的输入规划中学习动作代价的算法。我们提供了理论和实证结果,表明$LACFIP^k$能够成功解决这一任务。

英文摘要:

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.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑