Utilizing Support Vector Machine Algorithm and Feature Reduction for Accurate Breast Cancer Detection An Exploration of Normalization and Hyperparameter Tuning Techniques

Abstract

In this work, we will evaluate the impact of independent component analysis (ICA) on a breast cancer decision support system's feature reduction capabilities. The Wisconsin Diagnostic Breast Cancer (WDBC) dataset will be utilised to construct a one-dimensional feature vector (IC). We will study the performance of k-NN, ANN, RBFNN, and SVM classifiers in spotting mistakes using the original 30 features. Additionally, we will compare the IC-recommended classification with the original feature set using multiple validation and division approaches. The classifiers will be tested based on specificity, sensitivity, accuracy, F-score, Youden's index, discriminant power, and receiver operating characteristic (ROC) curve. This effort attempts to boost the medical decision support system's efficiency while minimising computational complexity.