Compound Decision Procedures for Pattern Classification

Download this report (PDF, 1.46 MB, 107 pages)
Report Number: AMRL TR 67-10
Author(s): Abend, Kenneth
Corporate Author(s): Philco-Ford Corporation
Laboratory: Biomedical Laboratory
Date of Publication: 1967-12
Pages: 107
Contract: AF 33(615)-2966
DoD Project: 7233 - Biological Information Handling Systems and Their Functional Analogs
DoD Task: 723305 - Theory of Information Handling
Identifier: AD0667570

Abstract:
Compound decision theory is shown to be powerful as a general theoretical framework for pattern recognition, leading to nonparametric methods, methods of threshold adjustment, and methods for taking context into account. The finite-sample-size performance of the Fix-Hodges nearest-neighbor nonparametric classification procedure is derived for independent binary patterns. The optimum (Bayes) sequential compound decision procedure, for known distributions and dependent states of nature is derived. When the states of nature form a Markov chain, the procedure is recursive, easily implemented, and immediately applicable to the use of context. A similar procedure, in which a decision depends on previous observations only through the decision about the preceding state of nature, can (when the populations are not well separated) yield results significantly worse than a procedure that does not depend on previous observations at all. When the populations are well separated, however, an improvement almost equal to that of the optimum sequential rule is achieved.

Provenance: RAF Centre of Aviation Medicine

Other options for obtaining this report:

Via the Defense Technical Information Center (DTIC):
A record for this report, and possibly a pdf download of the report, exists at DTIC

Via National Technical Report Library:
This report may be available for download from NTRL. Use the Title from this record to locate the item in DTIC Online

Indications of Public Availability
No digital image of an index entry indicating public availability is currently available
There has been no verification of an indication of public availability from an inside cover statement



Export