Hyperparameter-Optimized Machine Learning Techniques for Mammogram Classification

Abstract

Computer technology has employed Machine Learning models in a variety of applications to improve performance. The hyperparameter of a machine learning model must be adapted to overcome learning limitations and increase its performance. In this research, the hyperparameters of machine learning classifiers are tuned to identify cases of benign or malignant breast abnormalities. An experimental investigation was conducted using the Wisconsin Diagnosis Breast Cancer (WDBC) Dataset. A fusion model, Bayesian Optimization Hyper Band-Naïve Bayes (BOHB-NB) is employed, which is combined with conventional classification approaches like Logistic Regression (LR), Naive Bayes (NB), and Support Vector Machine (SVM).  The proposed methods are compared to cutting-edge models like SVM, NB, LR, K-Nearest Neighbour (KNN), Random Forest, and Decision Tree using a wide range of parametric measures, such as Precision, Recall, Specificity, F-measure, Accuracy, True Positivity Rate (TPR), and False Positivity Rate (FPR). The results show that the proposed methods outperform the leading models.