CHARACTERIZATION OF HUMAN BRAIN'S YURAGI (UCTUATION) IN NEURAL SIGNALS RECORDED NON-INVASIVELY

Abstract

Random noise typically described as undesirable disturbance has always been suppressed by spending large amount of energy whether it is the eld of electronics, computer tech- nology based arti cial intelligence or robotics. In contrast the e ect of noise on non- linear system like biological system has been found to be positive. Naturally existing noise in the environment is utilized by the biological system for their improved performance in infor- mation processing. Yuragi (in Japanese) is referred to as biological uctuation. Inspired from this biological activity of exploiting noise for seeking the potential bene ts of low power consumption, adaptive to the environment and stable, there has been signi cant progress in development of applications of Yuragi to robotic systems as a complementing work between robotics and biology. In order to develop the mathematical function of such bene cial noise that is biologically inspired its necessary to understand the struc- ture of physiological noise. This thesis tries to demonstrate a newly adopted paradigm to retrieve the structure of noise from neural recording of human brain. EEG dataset has been recorded from a single channel, dry electrode device. The brains electrical activity recorded non- invasively from the device is contaminated by various types of noise, both of biological and non- biological nature. This work tries to characterize both type of noise and intends to nd the emerging pattern from biological noise and reject non biological

uctuation which can potentially be from electrical interference present in environment and acquired while travelling through electrode.