A New Hybrid Model for Classifying Cardiac Arrhythmias Using Convolutional Neural Networks and Particle Swarm Optimization

To detect cardiac arrhythmias (CAs), a test that is not invasive called an electrocardiogram (ECG) is performed. Segmentation is essential for the diagnosis of cardiac arrhythmias. Various computational and mathematical techniques have been used to this categorization. In this article, we present a novel computer framework for the categorization of CAs using the MIT-BIH Arrhythmia Dataset (MITDB) using the particle swarm optimization (PSO) technique as well as a network of convolutional neural networks (CNN). The PSO seeks to maximize accuracy & minimize category cross entropy error (CE) by optimizing hyper parameters that determine a CNN's layers of structure. In both training and testing, the suggested model discovered a good structure of layers in 0.044968 & 0.084768 days, respectively. These findings prove the validity of the suggested approach & show that it is a novel method for training CNNs since it eliminates the need for a human to manually set hyper parameter values for the network's layers.