*College of Administrative Science, Department of Quantitative & Information Systems, Kuwait University, P.O. Box: 5486, Safat, 13055, Kuwait, e-mail: jafar@kuc01.kuniv.edu.kw 

**Microsystems Lab, electrical Engineering Department, University of Maryland, College Park, MD 20742, USA. 

ABSTRACT

In this paper, an algorithm for the design of functional link single layer neural networks using N binary inputs is described. The resulting connection weights are pure integer numbers.  These integer weights facilitate faster learning by the neural network due to binary operations rather than algebraic multiplications.  Comparison between various learning algorithms, mainly back propagation with the delta rule, and the functional link approach is demonstrated via several recognition applications.  It is illustrated that the functional link approach, due to enhancing input patterns, produces a robust algorithms for linearly non-separable classification problems in terms of processing speed and convergence.  Furthermore, implementation of the neural network can be accomplished using the vast availability of off-the-shelf components and VLSI techniques.

 Keywords: back propagation with the delta rule; functional link network; linearly non-separable classification; preceptron, recognition applications.