Bio-Inspired Ensemble Feature Selection (Biefs) and Ensemble Multiple Deep Learning (Emdl) Classifier for Breast Cancer Diagnosis

Abstract

Breast cancer is a major disease diagnosed in women, affecting 2.1 million women every year, and is the reason for most cancer-related mortality in women, as per the World Health Organization (WHO). For cancer researchers, accurately forecasting the life expectancy of breast cancer patients is a serious challenge. Bio-Inspired Ensemble Feature Selection (BIEFS) algorithm is introduced uses three feature selection methods such as Adaptive Mutation Enhanced Elephant Herding Optimization (AMEHO), Adaptive Mutation Butterfly Optimization Algorithm (AMBOA), and Adaptive Salp Swarm Algorithm (ASSA) and integrates their normalized outputs for getting quantitative ensemble importance. BIEFS algorithm depends upon the aggregation of multiple FS techniques by Pearson Correlation Coefficient (PCC). Ensemble Multiple Deep Learning (EMDL) classifier is introduced which combines several individual models such as AdaBoost-Convolutional Neural Network (A-CNN), Long Short-Term Memory Network (LSTM), and Deep Auto-Encoder (DAE) is introduced to obtain better generalization performance. Then, these three classifiers such as A-CNN, LSTM, and DAE are used for ensemble classification using a Weight Majority Voting (WMV) mechanism. Wisconsin Diagnosis Breast Cancer (WOBC), and Wisconsin Diagnosis Breast Cancer (WDBC) are collected from University of California, Irvine (UCI) repository for experimentation. Evaluation metrics like Precision, Recall, F-measure, Accuracy, and Area Under Curve (AUC) are used to compare the results of proposed system and existing classifiers. All experiments are executed within a simulation environment and conducted in MATrix LABoratory R 2014 a (MATLAB 2014a) tool. The proposed EMDL classifier is showing better performance as compared to the traditional classification models.