HMLP Network for ECG Classification Based Training Algorithms

A precise evaluation of heart function is required whenever patients display symptoms of abnormal heart rhythm. The study carries out training on algorithms to spot diseases. The electrocardiogram (ECG) signal shows important details that record the amplitudes and durations of the P, QRS and T waves as the input parameter. Modified version of multilayer perceptron (MLP) network, the hybrid multilayer perceptron (HMLP) network capable of improving prediction performance. Bayesian Regularization (BR), Levenberg Marquardt (LM) and Backpropagation (BP) training algorithms are applied in the research to train the MLP and HMLP networks. From the works, HMLP network trained by BR training algorithm capable to give better results by achieving 0.72 of mean square error (MSE) and 0.96 of regression.