Function approximator design using genetic algorithms

The approximation of a mathematical function (using examples in the form of input-output pairs) is a central issue in subjects as diverse as pattern recognition, control theory and statistics. In this paper, we propose an approach for designing a universal function approximator based on a combination of trigonometric and polynomial functions using genetic algorithms (GAs). We performed some experiments using our proposed approach and compared the results to several existing approaches. The results were promising in that the proposed approach was found to be superior to the other approaches in approximating a variety of test functions.