The application of a genetic approach as an algorithm for neural networks

Abstract

Using artificial neural nets (ANNs) should help to find solutions for problems that are difficult to handle by conventional algorithms (for example: pattern recognition or language processing). The problems are not coded directly by an algorithm. They are to be solved by constructing a neural net, which is capable of learning. Hence an important research area in artificial intelligence is the construction and trial of different learning algorithms. The learning approaches to neural nets are modified algorithms from optimization theory. The goal of this report is the presentation of a learning paradigm for neural networks, which is very different from the other learning approaches: learning by genetic algorithms.