
Accession Number : AD0720837
Title : Pattern Recognition with Continous Parameter, Observable Markov Chains.
Descriptive Note : Interim scientific rept. no. 10,
Corporate Author : MICHIGAN STATE UNIV EAST LANSING DIV OF ENGINEERING RESEARCH
Personal Author(s) : Dubes,Richard C. ; Panayirci,Erdal
Report Date : 25 NOV 1970
Pagination or Media Count : 34
Abstract : The paper develops Bayesian learning and decisionmaking algorithms for the following pattern recognition problem. Each of M pattern classes is described by a continuousparameter, discretestate Markov chain having a finite number of states. All states and times of transition between states can be observed perfectly. The transition rate matrices, which establish the properties of the chains, are not known a priori. A Bayesian learning algorithm using a fixed amount of memory digests the training patterns which consist of a member function from each chain. This leads to an iterative, computationally simple, decisionmaking algorithm for classifying any portion of a member function. The Bhattacharyya bound and the probability of error are derived for the 2state, 2chain problem when the transition rate matrices are known. The last section reports on a computer simulation of a 3state, 2chain problem with varying amounts of training data. An appendix summerizes the pertinent facts about Markov chains. (Author)
Descriptors : (*PATTERN RECOGNITION, *DECISION THEORY), STOCHASTIC PROCESSES, DECISION MAKING, ELECTROENCEPHALOGRAPHY, MATRICES(MATHEMATICS), ERRORS, ALGORITHMS
Subject Categories : Operations Research
Cybernetics
Bionics
Distribution Statement : APPROVED FOR PUBLIC RELEASE