PARALLEL PROCESSING AND IMPLEMENTATION OF NEURAL NETWORKS
Keywords:
neural networks, parallel processing, computer performance, von Neumann architecture, neural processors, pattern recognition, artificial intelligenceAbstract
This article provides a comparative analysis of the performance of the human brain and traditional sequential computers, and also substantiates the need to transition to parallel information processing. The authors examine in detail the technological limitations of creating fully-fledged neural computers, such as the interconnection problem, the three-dimensional nature of the connection structure, and the complexity of programming. The paper identifies alternative methods for modeling neural networks at the present stage and outlines the range of applied problems in which neural network technologies demonstrate the greatest effectiveness.
References
1. Уоссермен Ф. Нейрокомпьютерная техника: теория и практика. — М.: Мир, 1992.
2. Мкртчян С.О. Нейроны и нейронные сети. (Введение в теорию формальных нейронов). - М.: Энергия, 1971. - 232 с.
3. Гилл Ф., Мюррей У., Райт М. Практическая оптимизация. - М.: Мир, 1985.-509 с.
4. Лоскутов А.Ю., Михайлов А С. Введение в синергетику. - М.: Наука. Гл. ред. физ.-мат. лит., 1990. - 272 с.
