netGEN - A Parallel System Generating Problem-Adapted Topologies of Artificial Neural Networks by Means of Genetic Algorithms
Reinhold Huber, Helmut A. Mayer, Roland Schwaiger · 1995
Artificial neural networks (ANNs) have shown to perform satisfactorily for pattern recognition tasks. It has also been shown that ANNs are superior to some of the classical statistical methods in pattern classification, but little is known how to design the ANN. A genetic algorithm (GA) based method can be used to determine the ANN architecture for a specific task. We describe the netGEN system which is an existing implementation of a GA evolving ANNs in parallel. A simple pattern recognition task is solved so as to demonstrate the performance of netGEN. 1 Introduction We describe a parallel system for the pattern extraction task by the ANN approach. Compared to statistical classifiers, such as Bayesian a posteriori classifiers, ANN classifiers have the important characteristic that no underlying distributional form for the class densities is assumed [Lip93]. Due to the independency from a priori statistical parameters and their inherent parallel nature, we decided to use ANNs to solv...