Conference proceedings article
Kernel Based Chaotic Firefly Algorithm for Diagnosing Parkinson’s Disease


Research Areas
Currently no objects available

Publication Details
Author list: Sujata Dash, Ajith Abraham and Atta-ur-Rahman
Publication year: 2020
Title of series: Advances in Intelligent Systems and Computing
Volume number: 923
Start page: 176
End page: 188
Number of pages: 13
Web of Science ID:
PubMed ID:
Scopus ID: 85064883669
Languages: English-United States


Parkinson’s disease (PD) is prevalent all over the world and the amount of research carried out so far is not sufficient to precisely diagnose the disease. In this context, a novel kernel based chaos optimization algorithm is proposed by hybridizing a chaotic firefly algorithm (CFA) with kernel based Naïve Bayes (KNB) algorithm for diagnosing PD patients employing a voice measurement dataset collected from UCI repository. Six different chaotic maps are used to develop six CFA models and compared to find the best chaotic firefly model that can enhance the global searching capability and efficiency of CFA. To select the discriminative features from the search process, non-parametric kernel density estimation is used as a learning tool for the Bayesian classifier for evaluating the predictive potential of the features from four different perspectives such as size of the feature subset, classification accuracy, stability, and generalization of the feature set. The simulated results showed that logistic based CFA-KNB model outperformed other five chaotic map based CFA-KNB models. The generalization and stability of the predictive model is established by computing the model with four well-known representative classifiers viz., Naïve Bayes classifier, Radial Basis classifier, k-Nearest Neighbor, and Decision Tree in a stratified 10-fold cross-validation scheme. By finding appropriate chaotic maps, the proposed model could able to assist the clinicians as a diagnostic tool for PD.


Projects
Currently no objects available

Keywords
Currently no objects available

Documents
Currently no objects available

Last updated on 2019-05-05 at 14:18