摘要

We consider the problem of search the optimal by complexity discriminant function. The criteria of quality for discriminant functions are described, which are developed within the framework of the group method of data handling: criterion based on partition of observations into initial and testing samplings, and criterion of sliding test. The problems of this class belong to pattern recognition problems under the condition of structural uncertainty, which were considered by A.G. Ivakhnenko as long ago as 60−70-th of the last century as urgent problems of engineering cybernetics.