Prioritization Methods for Accelerating MDP Solvers
Prioritization Methods for Accelerating MDP Solvers
复制标题
加速 MDP 求解器的优先级划分方法
DOI：
--
复制
发表时间：
2005
影响因子：
6
通讯作者：
Kevin Seppi
中科院分区：
文献类型：
--
作者：
D. Wingate;Kevin Seppi
通过消除冗余或无用的备份，并以正确的顺序备份状态，可以显著提高值和策略迭代的性能。我们研究了几种旨在加速这些迭代求解器的方法，包括优先级，分区和变量重新排序。我们生成一个家庭的算法相结合的几种方法讨论，并提出了广泛的经验证据表明，性能可以提高几个数量级的许多问题，同时保持准确性和收敛性的保证。
The performance of value and policy iteration can be dramatically improved by eliminating redundant or useless backups, and by backing up states in the right order. We study several methods designed to accelerate these iterative solvers, including prioritization, partitioning, and variable reordering. We generate a family of algorithms by combining several of the methods discussed, and present extensive empirical evidence demonstrating that performance can improve by several orders of magnitude for many problems, while preserving accuracy and convergence guarantees.