Article details

Title: Genetic Optimal Hardware Implementation of a Feedforward Neural Network
Author(s): Constantin-Iulian Vizitiu               

Abstract: Genetic algorithms are an important class of modern global searching methods with specific applications in the complex optimization problems. The aim of this paper is to propose an interesting approach of a feedforward neural network topology optimization based on a new fitness function definition. The experimental results obtained in this case are compared with the ones from a classic pruning method, and for their validation, a hardware implementation is indicated.

Keywords: neural network, pruning technique, genetic algorithm, pattern recognition.

References:
[1] C.S. LINDSEY – Neural Networks in Hardware: Architectures, Products and Applications, Lectures, Royal Institute of Technology, Stockholm, Sweden, 2002
[2] S.K. PAL, P.P. WANG – Genetic Algorithms for Pattern Recognition, CRC Press, Boca Raton, FL, 1996
[3] C. VIZITIU – A Genetic Algorithm for Optimization of a Feedforward Neural Network Topology, Proc. of the 7th International Conference “Communications 2008”, pp. 173-176, Bucharest, Romania, Jun. 5-7, 2008
[4] C. VIZITIU – Contributions Concerning the Video Pattern Recognition with Neural Networks, Ph.D. Thesis, Military Technical Academy, Bucharest, Romania, 2003 (in Romanian)