Abstract
Various implementations of the functional basis of neural network classifiers are considered in this paper. Special attention is given to the implementation aspects of classifiers based on discriminant analysis, such as the Winner-Takes-All (WTA) competition mechanism and the ranking of classification results. It is proposed to determine the maximum value of the linear discriminant function using a sorting method with a high-speed decrement operation, while ranking is performed through a high-speed increment operation. The ranking mechanism aims to enhance the classification process through visualization and to facilitate decision-making, particularly in medical diagnostics, through the use of decision support subsystems.