The Analysis of Shooting Problems via Generalised Bandits
Abstract
A single Red wishes to shoot at a collection of Blue targets in order to maximize some measure of return obtained from Blues killed before Red's own demise. While the class of decision processes called multi-armed bandits has been previously deployed to develop optimal policies for Red, we argue the importance of a little known, but more general class of bandit processes introduced by Nash (1980). In particular, the deployment of this class of processes will enable Red to take account in a natural way of the relative threats posed to his own survival in taking targeting actions. We develop optimal shooting policies for Red in the context of a range of models, which are of independent interest. The paper concludes with a numerical study.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2004
- Accession Number
- ADA424812
Entities
People
- Donald P. Gaver Jr.
- Helen M. Mitchell
- Kevin D. Glazebrook
- Patricia A. Jacobs
Organizations
- Naval Postgraduate School