Automatic Sleep Stage Classification Using a Taguchi-Based Multiscale Convolutional Compensatory Fuzzy Neural Network

Current methods for sleep stage detection rely on sensors to collect physiological data. These are inaccurate and take up considerable medical resources. Thus, in this study, we propose a Taguchi-based multiscale convolutional compensatory fuzzy neural network (T-MCCFNN) model automatically detect classify stages. In the proposed T-MCCFNN model, convolution kernels extract features of input electroencephalogram signal is used place traditional fully connected as classifier improve convergence rate during learning reduce number parameters required. Due complexity general deep networks, trial error often determine their parameters. However, method very time-consuming. Therefore, study uses Taguchi instead, where optimal parameter combination identified over minimal experiments. We use Sleep-EDF database evaluate model. The results indicate that classification accuracy 85.3%, which superior by other scholars.