ECG Classification Performing Feature Extraction Automatically Using a Hybrid CNN-SVM Algorithm

Özaltın Ö., Yeniay M. Ö.

IEEE, 3rd International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA), vol.2021, no.10.1109/HORA52670.2021.9461295, pp.1-5, 2021 (Scopus)


In this study, we have presented a hybrid Convolution Neural Network (CNN)-Support Vector Machine (SVM) algorithm which has overcome overfitting for classifying Electrocardiogram (ECG) signals that have been transformed to 2D images using continuous wavelet transform (CWT). We also have suggested ProposedNet that is a kind of convolutional neural network algorithm. Also, it has been trained more than once. Moreover, it has performed feature extraction automatically. We have compared the ProposedNet which has 34 layers, with SVM. Additionally, we also have compared ProposedNet-SVM that is also our suggestion, with these algorithms. Comparison results indicate that ProposedNet, SVM, and ProposedNet-SVM have been achieved accuracy rates of 95.6%, 89.17%, and 99.524% respectively.