UDC 004.855.5
SEARCH ALGORITHM OF THE PARAMETERS VALUES OF THE BSMOTE-ALGORITHM IN THE PROBLEM OF THE SVMCLASSIFICATION BASED ON THE IMBALANCED DATASETS
L. A. Demidova, PhD (technical sciences), full professor, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it.
I. А. Klyueva, post-graduate student, RSREU, Ryazan; This email address is being protected from spambots. You need JavaScript enabled to view it..
The problem of SVM (Support Vector Machine) classification based on the unbalanced data sets applied to generate train sets, using the synthetic algorithm sampling – bSMOTE algorithm (borderline Synthetic Minority Oversampling Technique algorithm) has been considered. The aim of this work is the development of the search algorithm of bSMOTE-algorithm parameters values in the problem of SVM classification of unbalanced datasets, providing the reducing of time expenditures for the development of SVM classifier, characterized by high quality of data classification. The search of SVM classifier parameters values has been implemented with the use of PSO algorithm (Particle Swarm Optimization algorithm). The results of experimental studies confirming the feasibility of the search algorithm of bSMOTE-algorithm parameters values in the problem of SVM classification of unbalanced datasets have been given.
Key words: imbalanced data, sampling, bSMOTE-algorithm, classification, SVM classifier, radial basis kernel function, PSO algorithm.