Formation of Fuzzy Patterns in Logical Analysis of Data Using a Multi-Criteria Genetic Algorithm

The formation of patterns is one the main stages in logical data analysis. Fuzzy approaches to pattern generation analysis allow cover not only objects target class, but also a certain proportion opposite class. In this case, search an optimization problem with maximum coverage class as objective function, and some allowed constraint. We propose more flexible symmetric model which does impose strict restriction on observations. Instead, our converts such (purity restriction) into additional criterion. Both, are two functions problem. for balance these criteria essence proposed method. modified evolutionary algorithm based Non-dominated Sorting Genetic Algorithm-II (NSGA-II) solve new uses approximation Pareto set considers solution’s representation informativeness patterns. have tested approach applied medical problems classification under conditions sample asymmetry: significantly dominated other. results were comparable and, cases, better than commonly used machine learning algorithms terms accuracy, without losing interpretability.