Abstract:
Lie is one form of deception. People wants to differentiate between truth and lie but it is
not always possible. Truth always show the right path. From ancient time, different
methods were applied to find out the truth. Mostly in judiciary and for security of a nation,
we need to check whether a person is telling the truth or not. Most of the methods are based
on body’s states that can be altered easily. But brain waves are not easy to alter. So, in this
research project we have tried to detect lie by analyzing brain waves. Brain waves can be
detect using EEG Headset. It produces lots of data of brain signal that has been analyzed
using data mining techniques. And machine learning & neural network algorithms are used
to make decision based on the collected data. Neural oscillation are collected using an EEG
headset during question-answer session. After that data were preprocessed and then three
algorithms (two class locally deep SVM, two class decision forest tree and two class neural
network) that were selected were applied on the data. We have found 63.1% accuracy in
Two Class locally Deep SVM, 99.2 % accuracy in Two Class Decision Forest and 55.7%
accuracy in Two Class Neural Network algorithm.