Title :
Hybrid pattern recognition using Markov networks
Author :
Gregor, J. ; Thomason, M.G.
Author_Institution :
Inst. of Electron. Syst., Aalborg Univ., Denmark
fDate :
6/1/1993 12:00:00 AM
Abstract :
Markov networks are inferred automatically for different classes of learning strings. In subsequent string-to-network alignments for test samples, the networks are used to deduce structural characteristics and to provide similarity measures. By processing the similarity measures as numerical-value features, standard nonparametric decision-theoretic pattern classifiers may be applied to determine class membership. The nearest-neighbor rule and linear discriminant-function classifiers are discussed, and their performances are compared with that of a maximum-likelihood classifier. The hybrid system´s ability to determine string orientation correctly is investigated. Experiments with several thousand human banded chromosomes are reported
Keywords :
Markov processes; biology computing; decision theory; nonparametric statistics; pattern recognition; Markov networks; biology computing; class membership; human banded chromosomes; hybrid pattern recognition; learning strings; linear discriminant-function classifiers; maximum-likelihood classifier; nearest-neighbor rule; nonparametric decision-theoretic pattern classifiers; similarity measures; string orientation; string-to-network alignments; structural characteristics; Biological cells; Computer networks; Computer science; Computer vision; Humans; Liquid crystals; Markov random fields; Optimized production technology; Pattern analysis; Pattern recognition;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on