Reuse of program trees in genetic programming with a new fitness function in high-dimensional unbalanced classification

Abstract

Genetic programming (GP) may also evolve biased classifiers when having the class imbalance issue. Class imbalance is a difficult but important issue, and high-dimensionality brings difficulty when addressing the class imbalance issue. This paper focuses on addressing the performance bias of GP in classification with high-dimensional unbalanced data, with the goal of increasing the accuracies of the majority class and the minority class, as well as saving the training time. In this paper, a new fitness function is developed to address the class unbalanced issue, and moreover, a strategy is proposed to reuse previous good GP trees when using multiple GP processes to build a multi-classifier system. Experimental results show the better performance of the proposed method.