On Inductive Construction of the Prospector–Like Knowledge Bases

Abstract

The problem of inductive construction of the Prospector–like knowledge bases in the form of weighted if–then rules is considered. The modification of two existing realization schemes of the method of structural risk minimization for an automatic search of the optimal classifiers family is proposed. The experiments on the training sets of UCI repositaries data bases show that this modification is reasonable and applicable well with good results.