Combining modular neural networks developed by evolutionary algorithm

Evolutionary approach to artificial neural networks has been rapidly developing in the recent years and shows a great possibility as a powerful tool. However, most evolutionary neural networks use the simple node as a building block to evolve, and select the only one network producing the best result after evolution. In this paper we present the concepts and methodologies for evolutionary modular neural networks which boost up the overall performance by combining several potential networks emerged in the course of evolution. The experimental result with the recognition problem of handwritten numerals shows the possibility of combining a number of characteristic networks from gene pool.