Performance Analysis of EEG Signals using Conventional and Hybrid Artificial Neural Network

Brain Computer Interface (BCI) using Electrooencephalogram (EEG) is one of the versatile tools to measure the brain thoughts and convert them to operate the external devices in the deficiency of biological controls. These techniques are used to develop the rehabilitative devices for the individual person who affected with locked in syndrome (LIS). The main reason for LIS is due to death of motor neurons. To overcome the real world problem we conduct our experiment with eight normal patients for four tasks using three electrode system and ADI T26 bio amplifier with Lab chart. Four task signals are applied to statistical method to retrieve the twenty two features and trained with the feed forward neural network (FFNN) and feed forward neural network with wolf Grey optimization algorithm (FFNNWGOA) to see network model which was more perfectly supported to identify the tasks. The study showed that statistical features through feed forward neural network with grey wolf optimized algorithm classifier produced maximum performances of 94.06% compared to other feed forward neural network classifier model and also we identified that optimized features demonstrated maximum performances in minimum time duration during training in all the following twenty trials.