Neural Networks and Fuzzy-Logic Control on Personal Computers and Workstations

Most neural-network programs for personal computers and engineering workstations simply control a fixed set of canned network-layer algorithms with pulldown menus. This hands-on tutorial demonstrates both neural networks and fuzzy-logic control with a different approach. A natural, computer-readable notation for matrix operations and differential equations lets users create their own neural networks and fuzzy-logic controllers on the screen; real simulation experiments then execute immediately.

Neural Networks and Fuzzy-Logic Control introduces a simple integrated environment for programming displays and report generation. It includes the only currently available software that permits combined simulation of multiple neural networks, fuzzy logic controllers, and dynamic systems such as robots or physiological models. The enclosed educational version of DESIRE/NEUNET differs for the full system mainly in the size of its data area and includes two screen editors, compiler, color graphics, amd many ready-to-run examples. The software lets users or instructors add their own help screens and interactive menus.

Differential equations in scalar and/or matrix form are entered in ordinary mathematical notation. Users can program new neural networks or fuzzy logic applications without learning C or LISP. For truly interactive experimentation, the extra-fast compilation is unnoticeable, and simulation speed still compares well with that of special accessory processors. The 123 figures include block diagrams, simulation-output graphs produced on personal computers and engineering workstations and many complete computer programs.

The version of DESIRE/NEUNET included here is for PCs, viz. 286/287, 386/387, 486DX, Pentium, P6, SX with math coprocessor.

A Bradford Book

Granino A. Korn has been a Professor of Electrical Engineering at the University of Arizona.