Novel classifiers for intelligent disease diagnosis with multi-objective parameter evolution. (April 2018)

• Record Type: Journal Article
• Authors: MadhuSudana Rao, Nalluri; Kannan, Krithivasan; Gao, Xiao-zhi; Roy, Diptendu Sinha
• Is Part Of: Computers & electrical engineering. Volume 67(2018)
• Page Start: 483
• Page End: 496
• Publication Date: 2018-04
• Subjects: Medical engineering data -- Disease classifier -- Data mining -- Competitive collection -- Sequential Minimal Optimization (SMO) -- Support Vector Machine (SVM) -- Initial population generation -- EHO -- MOEA/D -- NSGA-II; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2018.01.039
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: In this research, intelligent classifiers for disease diagnosis are designed that use classifier parameters, such as cost, tolerance, gamma and epsilon, with multi-objective evolutionary algorithms. The multiple objective functions are prediction accuracy, sensitivity and specificity. This paper employs a Sequential Minimal Optimization (SMO), a variant of the classical Support Vector Machine (SVM), as the base classifier in conjunction with three popular evolutionary algorithms (EA), namely, Elephant Herding Optimization (EHO), Multi-objective Evolutionary Algorithm Based on Decomposition (MOEA/D) and Non-dominated Sorting Genetic Algorithm-II (NSGA-II), for parameter evolution. A new cuboids based initial population generation mechanism was also introduced to hybridize EHO, called CEHO. The performance of CEHO is compared with the other three EAs (EHO, MOEA/D and NSGA-II) over 17 medical engineering datasets, and pertinent statistical tests were conducted to substantiate their performances. The results demonstrate that the proposed CEHO exhibit better to competitive results across all datasets.