Problem-oriented parallel processors for the solving of classification tasks

Abstract

This paper describes in detail, the workings of a prototype version of a parallel processor for high-performance computing of large-scale tasks for identification (classification). The use of such processors will provide the required computing power with a guaranteed quality of decision making. Program settings and architecture cascading, as well as the knowledge of the dependence of performance of hardware resources, will reduce the time frame for the development of intelligent systems for pattern recognition as integrated software and hardware, presented in real-time.