ADAPTIVE MODELLING OF LIKELIHOOD CLASSIFICATION-II.
Final rept. 4 Apr 66-23 Nov 67,
PHILCO-FORD CORP BLUE BELL PA ADVANCED ENGINEERING AND RESEARCH LAB
Pagination or Media Count:
The report is largely devoted to compound decision procedures and their application to pattern classification applications presented include taking context into account, the combatting of intersymbol interference adaptive threshold adjustment, and the classification of non-stationary time series. A multivariate, two-sample, rank-order classification procedure is described. The characterization of the complexity of pattern recognition problems and recognition systems is considered. Computer experiments, comparing a self-adjusting threshold procedure with a Bayesian procedure, are also presented. Author