Enhanced Breast Cancer Classification through Data Fusion Modeling

Abstract

This study addresses issues of classifier instability and poor adaptability to sample distribution in intelligent breast cancer diagnosis. We propose a novel classifier construction algorithm based on Adaboost, integrating BP, RBF, and Naïve Bayes networks. Firstly, multiple weak classifiers are trained using different classification algorithms. Subsequently, a weight allocation strategy is employed, increasing the weight of misclassified diseased samples as healthy and decreasing the weight of misclassified healthy samples as diseased during data distribution processing. Finally, the adjusted weights are used to recombine the weak classifiers into a strong classifier. Experimental validation on the Wisconsin Breast Cancer (WBCD) dataset from the UCI (University of California, Irvine) database demonstrates the superiority of the proposed classification model over individual algorithms. This algorithm's application is expected to enhance the accuracy and stability of breast cancer diagnosis, providing valuable insights for the further development of intelligent diagnostic systems.