An Aggregation Procedure for Large-Scale Markov Decision Processes

被引：0

作者：

Bartl, Ondrej ^{[1
]}

机构：

[1] Univ Zilina, Fac Management Sci & Informat, Dept Software Technol, Zilina 01026, Slovakia

来源：

PROCEEDINGS OF THE 22ND INTERNATIONAL CONFERENCE ON MATHEMATICAL METHODS IN ECONOMICS 2004 | 2004年

关键词：

Markov/semi-Markov decision processes; state and action aggregation;

D O I：

暂无

中图分类号：

F [经济];

学科分类号：

02 ;

摘要：

Markov decision models with a high state space cardinality may resist being computationally tractable. Then reduction in the number of possible state variable values can help. The fixed-weight aggregation procedure for large-scale Markov/semi-Markov decision processes is described in the paper. A possibility to approximate an original decision model by action aggregation accompanying state aggregation is mentioned as well.

引用

页码：9 / 15

页数：7