Accession Number:

ADA336059

Title:

JPRS Report, Science & Technology, China, Fuzzy Logic & Neural Networks.

Descriptive Note:

Corporate Author:

JOINT PUBLICATIONS RESEARCH SERVICE ARLINGTON VA

Personal Author(s):

Report Date:

1991-02-27

Pagination or Media Count:

29.0

Abstract:

CMOS current mode circuit units are designed and fabricated completing various fuzzy logic operations and relevant processing. Experimental results show that these basic circuits have the advantages of simple structure, high functional density and high speed. They can be used as building blocks to achieve. VLSI implementation of fuzzy hardware. By the use of these circuit units a high speed fuzzy logic microprocessor for a real time hardware expert system has been designed. Back propagation rule has been shown to be an efficient learning algorithm for multilayered neural network. However it is limited because it only finds local minima. Roltzmann machine has also been shown to be an efferent learning rule. But it is limited because it learning rate is too slow. In this paper, we proposed and simulated a quantum learning algorithm for multilayered neural network. It is shown that its learning rate is more rapid than that of Boltzmann machine, and it can find the global minimum unlike back propagation algorithm does.

Subject Categories:

  • Government and Political Science
  • Anatomy and Physiology
  • Computer Programming and Software

Distribution Statement:

APPROVED FOR PUBLIC RELEASE