Fuzzy Logic on Long Short-Term Memory for Smart Person-Identification System through Electroencephalogram

Brain functions through communication in between the network of numerous nerves by means of electrical signals produced as a result of post synaptical potential. Electroen-cephalogram (EEG) is a technique in which specialized electrodes are placed in certain parts of the scalp in order to measure the magnetic fields arisen by the electrical signal transmission within the brain. The brain waves although follow a certain regular pattern, it still varies at a minute scale from person to person. This gives us the motivation to develop a smart decision-making system to identify a person based on brain waves. For this, we have developed a deep learning recurrent neural network model based on Long Short-Term Memory to outperform all existing work. We have also compared the results with other popular machine learning algorithm. The experiment has been conducted on 28 human subjects under different psychological conditions with consent and obtained over 90% identification accuracy. The model could be easily deployed for several theft or privacy protection application such as embedding the setup within a smart helmet and granting access to only the owner.