Shadow Gene Guidance: A Novel Approach for Elevating Genetic Programming Classifications and Boosting Predictive Confidence

This paper introduces a novel classification method that utilizes genetic programming (GP). The primary purpose of the proposed method is to enhance future generations of GP, through continuously refining the genetic makeup of the population for improved classification results. Accordingly, this paper developed the novel method by modifying Boruta feature selection method in such a way that allows to evaluate the significance of individuals' genes. This method creates modified versions of the genes called "shadow genes", evaluates their impact on model performance in competing with shadow genes, and identifies key genes. These key genes are then used to enhance future generations. The obtained results demonstrated that the proposed method not only enhances the fitness of the individuals but also steers the population toward optimal solutions. Furthermore, empirical validation on multiple datasets reveals that the proposed method significantly outperforms classic GP models in both accuracy and reduced prediction entropy, showcasing its superior ability to generate confident and reliable predictions.