AN ECG SIGNAL BASED FEATURE SELECTION FOR DYSRHYTHMIA CLASSIFICATION. USING PSO, GWO AND SVM

      Аннотация: Arrhythmia occurs when there is no proper working of electrical impulses present in the heart. An earlier detection of irregular heart rhythm is necessary in order to rescue ones survival. Classification of arrhythmia is needed for diagnosis. This report confers the Principle component analysis as feature reduction process to reduce high dimensional input without influencing classification methods and two feature selection techniques such as Grey wolf optimizer (GWO), Particle swarm optimization (PSO), and Support Vector Machine (SVM) helpful in choosing features with arrhythmia and resultswill be used for classification of various arrhythmia. Performance Analysis for these feature selection techniquesis estimated. The curse of dimensionality (i.e., dataset containing large volume of features) is solved using these feature selection methods. The result explores the performance metrics for integration of three methods such as PSO, GWO with SVO and shows that PSO and GWO integrated with SVM selected features with 96.08% accuracy.

Ключевые слова: Heart rate or rhythm, Cardiac arrhythmia, Classification of Arrhythmia, Feature reduction, Principle component analysis, Feature Selection, classification, Grey wolf optimizer, particle swarm optimization, Support Vector Machine