For the attention of readers!

摘要

In 2008 new books of prominent Ukrainian scientists in the field of informatics and computer engineering were issued.rnThe basis for Bayesian approach is determination of such structures of objects description for which the construction of efficient (optimal) recognition procedures is feasible. At present these matters have been solved for discrete objects with independent attributes and objects described by Markovian chains. There have been obtained deterministic estimates of errors of Bayesian recognition procedure depending on class dimensions (learning samples), a number of attributes and a number of values of attributes. The constructed estimates are essentially different from estimates of minimization methods of empirical risk. The monograph considers application of Bayesian recognition procedures to prediction of secondary protein structure, forecasting a dynamics of stock price.