[Home] [QCL] [NeuroGen] [WaveTools] [Programs] [Papers]

Genetic Algorithms for the Training of Neuronal Networks on Distributed Memory Systems

Current version: neurogen-1.0

The use of both, genetic algorithms and artificial neural networks, was originally motivated by the astonishing success of these concepts in their biological counterparts. Despite their totally different approaches, both can merely be seen as optimisation methods which are used in a wide range of applications, where traditional methods often prove to be unsatisfactory.

NeuroGen implements a hybrid parallel algorithm using genetic search and standard backpropagation to train neural networks on a transputer cluster. The package includes several test programs with can either be compiled to run on a transputer network or sequentially as a single threaded C programm.

Download

The current version of NeuroGen is 1.0:

Source Distriution: neurogen-1.0
Binary Distriution: neurogen-1.0-bin (ELF, i386, Linux 2.2, glibc2.1)

NeuroGen is written in ANSI C using only standard UNIX libraries and include files except for the parallel versions, which also use the Meiko Computing Surface Network library CS-Tools with their according header files for transputer specific functions.
Since I've written this programs during my ERASMUS semester in Newcastle and I don't have access to a transputer cluster with the above libs here, I can neither provide a binary package nor maintain plattform specific code for this plattform.

Documentation

Genetic Algorithms for Neural Network Training on Transputers · [Postscript] · [PDF]
The final report about my project on the use of genetic algorithms for the training of neuronal networks on distributed memory systems at the Department of Computing Science at the University of Newcastle.
This paper contains a detailed description of all algorithms and test programs as well as documentation on the source code.