An effective classification approach to categorize the breast cancer using modified support vector machine as a classifier

Abstract

AbstractBreast cancer is a type of cancer which threatens the health of women and occupies first in female cancer mortality. In recent days, the machine learning techniques are widely utilized in various medical fields particularly in the application related to detection and classification. However, the accuracy obtained from the existing machine learning techniques was reliably low and results in poor classification. To overcome the issues in existing algorithms, this research introduced a Modified Support Vector Machine (MSVM) to classify the breast cancer as benign and malignant. The input data is obtained from Wisconsin Breast Cancer Dataset (WBCD) and the pre‐processing is performed to remove the inappropriate information from the raw data. Then, Principle Component Analysis (PCA) is used in the process of dimensionality reduction. The reduced dimensions are subjected into the stage of feature extraction using linear Discriminant analysis (LDA) and finally the classification is performed using the proposed MSVM. The experimental results show that the proposed approach achieved better classification accuracy of 99.34% which is comparatively higher than the existing Deer canid based Deep Convolutional Neural Network (DCNN), LDA‐SVM, Fuzzy rule based system and Intelligent Ensemble Classification method on the basis of Multi‐Layer Perceptron (IEC‐MLP) with classification accuracy of 98.12%, 97.65%, 97.22% and 98.74% respectively.