Methods of automated classification based on wavelet-transform for automated medical diagnostics

Investigation into the methods of classification with training for determination of separating surfaces coefficients using optimization methods based on wavelet-transform with and without constraints (in the form of inequalities) is carried out. Recommendations are offered for choice of optimization method for classification in automated medical diagnostics systems with regard to the peculiarities of the phase in which classification is performed. These recommendations make it possible to justify the choice of versions of the method for the optimum search at classification depending on the stage of training (additional training) of the automated system and, as a rule, conflicting requirements on speed and accuracy of classification in a particular.