A New Class of Policies in Vector-Valued Markov Decision Processes

✍ Scribed by Kazuyoshi Wakuta

• Publisher: Elsevier Science
• Year: 1996
• Tongue: English
• Weight: 108 KB
• Volume: 202
• Category: Article
• ISSN: 0022-247X
• DOI: 10.1006/jmaa.1996.0337

✦ Synopsis

For a vector-valued Markov decision process with discounted reward criterion, we introduce a new class of policies called the semi-stationary policies and show that an optimal semi-stationary policy that attains the extreme points of the set of rewards induced by all policies can be described as a combination of optimal stationary policies.