Ensemble of support vector machines for heartbeat classification

This paper proposes a novel and simple method that uses the randomness of random matrix and SVM ensemble learning to discriminate eight types of heartbeats. We use random matrices to generate 15 groups of random features. Then we construct one SVM classifier on each group of random features along with a RR interval. The type of heartbeat is determined using majority voting strategy to combine 15 SVM classifiers. 3062 heartbeats obtained from the MIH-BIH electrocardiogram (ECG) database are used for experiments. The results show that our proposed method has an accuracy of 98.65% and is an effective method for heartbeat classification.