Гибридный метод распознавания на основе оптимальных тупиковых нечетких тестов и реккурентного метода наименьших квадратов

The hybrid method of recognition based on a combination of logic and statistical approaches to the data analysis and focused on solving recognition problems for multidimensional dynamically changing objects and problems with any size, including small, of learning sample is offered. Method is based on adaptive formation of the new description sufficient for construction of a reliable decision rule on the submitted sample of the data. The new description is formed by means of the clusters or syndromes construction with use of the optimal irreducible fuzzy tests and their coding with help of the least squares method.