Handling the problem of code bloating to enhance the performance of classifier designed using genetic programming

Abstract

The concept of "bloat" in Genetic Programming (GP) is a well established phenomenon characterized by variable-length genomes gradually increasing in size during evolution. "Bloating makes genetic programming a race against time, to find the best solution possible before bloat puts an effective stop to the search." In this paper we are proposing a special type of crossover operation in order to reduce the problem of bloat. In proposed crossover we are using local elitism replacement in combination with depth limit and size of the trees to reduce the problem of bloat substantially without compromising the performance. To demonstrate our approach we have designed a Multiclass Classifier using GP by taking few benchmark datasets. The use of local elitism in crossover increases the accuracy of the crossover operation and considering the fitness, depth limit and size of the trees after performing the elitism reduces the problem of bloat and further improves the performance of the classifier designed.