Program simplification in genetic programming for object classification

Abstract

This paper describes a program simplification approach in genetic programming (GP) to the use of simple algebraic techniques, prime numbers and hashing techniques for object classification problems. Rather than manually simplifying genetic programs after evolution for interpretation purpose only, this approach automatically simplifies genetic programs during the evolutionary process. This approach is examined on four object classification problems of increasing difficulty. The results suggest that the new simplification approach is more efficient and more effective than the basic GP approach without simplification. © Springer-Verlag Berlin Heidelberg 2005.