Deep learning for sleep stages classification: modified rectified linear unit activation function and modified orthogonal weight initialisation

Each stage of sleep can affect human health, and not getting enough at any may lead to disorder like parasomnia, apnea, insomnia, etc. Sleep-related diseases could be diagnosed using Convolutional Neural Network Classifier. However, this classifier has been successfully implemented into classification systems due high complexity low accuracy classification. The aim research is increase the reduce learning time proposed system used a modified Orthogonal Adam optimisation technique improve gradient saturation problem that occurs sigmoid activation function. uses Leaky Rectified Linear Unit (ReLU) instead function as an called Enhanced Sleep Stage Classification (ESSC) six different databases for training testing model on stages. These are University College Dublin database (UCD), Beth Israel Deaconess Medical Center MIT (MIT-BIH), European Data Format (EDF), EDF Extended, Montreal Archive Studies (MASS), Heart Health Study (SHHS). Our results show does exist anymore. optimiser helps noise which in turn result faster convergence time. speed ESSC increased along with better compared state art solution.