Multiclass maximum margin clustering via immune evolutionary algorithm for automatic diagnosis of electrocardiogram arrhythmias

Maximum margin clustering algorithm can obtain outstanding clustering performance by finding the maximum margin hyperplanes between clusters that can separate the data from different classes in an unsupervised way. However, it is only suitable for the clustering of small data set, since requires solving non-convex integer problem, which is computationally expensive. In this paper, to further improve the clustering performance, a new multiclass clustering method based on maximum margin clustering algorithm and immune evolutionary algorithm (IEMMMC) is proposed for diagnosis of electrocardiogram (ECG) arrhythmias. Five types of ECG arrhythmias obtained from MIT-BIH database are analyzed in the experiment, including normal sinus rhythm (N), premature ventricular contraction (PVC), atrial premature contraction (APC), fusion of ventricular and normal beat (FVN), fusion of paced and normal beat (FPN). And three types of performance evaluation indicators are used to assess the effect of the IEMMMC method for ECG arrhythmias, such as sensitivity, specificity and accuracy. Compared with K-means, fuzzy c-means and LS-SVM algorithms, our method reflects better performance not only in clustering result but also in terms of global search ability and convergence ability, which proves its effectiveness for the detection of ECG arrhythmias.