DWD-SVM: Support Vector Machines Based on the Distance Weighted Discrimination (DWD) Loss Function to Breast Cancer Diagnosis Classification
摘要
Breast cancer is the most prevalent invasive cancer among women and the second leading cause of cancer-related deaths in this population. It is generally classified into two categories: benign and malignant. In recent years, experts have emphasized addressing this critical issue. Machine learning advances have facilitated disease prediction and the prevention of life-threatening conditions. Numerous breast cancer prediction models have been developed using various statistical and machine learning techniques, leveraging datasets to build predictive models or extract valuable insights. This research used machine learning techniques on the Wisconsin Breast Cancer Dataset (WBCD) to develop an advanced diagnostic model. Specifically, we proposed a smooth support vector machine (SVM) framework incorporating the Distance Weighted Discrimination (DWD) loss function. This DWD-based approach offers several advantages over traditional SVM, including improved handling of imbalanced datasets, enhanced class separation, and greater robustness against outliers. Additionally, the resulting optimization problem was efficiently addressed using the Adam algorithm. Experimental results demonstrated that our method achieved an impressive accuracy of