DECISION RULES FOR REACTING TO NOISY SIGNALS FROM A WARNING SYSTEM.
Abstract
Using the minimization of total expected discounted cost as the criterion for optimality, this study investigated via dynamic programming, a class of stochastic sequential decision problems known as warning problems. Specifically, the study investigates two types of warning systems (Type I and Type II) with single and multiple signals. The dynamic programming models of Type I for single and multiple signals yielded a critical waiting time rule (CWT) as well as a critical probability value (CPV) rule as an optimal decision rule. However, for Type II, with two signals, the optimal rule is a CPV rule, but the corresponding waiting time rule, in general, is shown a CWT rule. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1970
- Accession Number
- AD0705522
Entities
People
- T. Y. Hans Tjian
Organizations
- University of California, Berkeley