Розробка ефективного класифікатора для автоматизованої системи допуску персоналу на виробництві

Abstract

A method of increasing the efficiency of the SVM-type classifier for the recognition of person's voice patterns in the automation system of personnel admission in production is proposed. Improving the quality of work of the classification algorithm was realized due to the determination of surface points in clusters of speakers by the proposed method of surface tension at the stage of training the recognition system.