arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2401.10431cs.AI
收藏

通过重放组合问题的解来学习 Monte Carlo Search 的先验

Learning a Prior for Monte Carlo Search by Replaying Solutions to Combinatorial Problems

LAMSADE, Université Paris Dauphine - PSL, CNRS(巴黎多菲纳大学-PSL,国家科学研究中心LAMSADE实验室)

Tristan Cazenave

AI总结:

本文提出利用已解组合问题的统计信息自动学习 Monte Carlo Search 先验,以替代手工启发式,在无推演时开销的同时显著提升性能,并验证于三类组合问题。

摘要:

Monte Carlo Search 在多个困难组合问题上给出了出色结果。在搜索过程中使用先验来执行非均匀推演,相较于均匀推演可大幅提升结果。针对组合问题定制的手工启发式方法常被用作先验。我们提出一种自动计算先验的方法。该方法利用已解决问题的统计信息,是一种简单且通用的方法,在推演时不会产生计算开销,并能带来显著的性能提升。该方法被应用于三个困难组合问题:Latin Square Completion、Kakuro 和 Inverse RNA Folding。

英文摘要:

Monte Carlo Search gives excellent results in multiple difficult combinatorial problems. Using a prior to perform non uniform playouts during the search improves a lot the results compared to uniform playouts. Handmade heuristics tailored to the combinatorial problem are often used as priors. We propose a method to automatically compute a prior. It uses statistics on solved problems. It is a simple and general method that incurs no computational cost at playout time and that brings large performance gains. The method is applied to three difficult combinatorial problems: Latin Square Completion, Kakuro, and Inverse RNA Folding.

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。

↑