CMVHHO-DKMLC: A Chaotic Multi Verse Harris Hawks optimization (CMV-HHO) algorithm based deep kernel optimized machine learning classifier for medical diagnosis

Abstract

In this manuscript, a Chaotic Multi Verse Harris Hawks Optimization algorithm based Deep Kernel Machine Learning Classifier (CMVHHO-DKMLC) method is proposed for medical diagnostics. By using CMVHHO, the feature selection (FS) is done for finding ideal feature subset of medical documents. In the proposed method, to obtain the ideal subset of characteristics for best classification, initially chaotic multi-verse optimization (CMV) was implemented to create first positions, after that Harris Hawks optimization (HHO) was utilized to keep positions informed current population at DKMLC-based search space. Then the proposed model is performed on MATLAB and the performance of proposed method is evaluated with assessment metrics. For demonstrating the efficacy of the proposed system on medical diagnosis, this manuscript carries out a series of comparative studies by testing with two medical data sets, i.e. Wisconsin Breast Cancer Database [40] and PIMA Indian Diabetic Dataset [41]. This proposed CMVHHO-DKMLC classifier provides 1.165% and 0.667% higher accuracy value compared with existing classifier model like against the chaotic multi-swarm whale optimizer-boosted support vector machine (CMWOAFS-SVM) and improved gray wolf optimization-based feature selection wrapped kernel extreme learning machine (IGWO-KELM) for breast cancer diagnosis. Similarly the proposed CMVHHO-DKMLC classifier provides 25.641% and 3.55% higher accuracy value compared with existing classifier model like chaotic multi-swarm whale optimizer-boosted support vector machine (CMWOAFS-SVM) and multi-objective firefly and multi-objective imperialist competitive algorithm optimizer-boosted support vector machine (MOFA-MOICA-SVM) for diabetic diagnosis. Simulation outcomes have demonstrated the dominance of proposed technique over other two competitive methods.