An Improved Approach for EEG Signal Classification using Autoencoder

Brain signals were started to use in deception detection process from last few years. Electroencephalogram (EEG) signals can reveal many important features of our thought which make it as a better tool for deception detection. A number of experiments were done in terms of visual stimuli based EEG signals. The purpose of this paper is to improvise the existing methods in the classification of familiar and unfamiliar faces which can be used as a basic model in deception detection. In this paper, we proposed a deep learning based classification of EEG signals for the given visual stimuli. In this experiment, the subjects were shown by familiar and unfamiliar faces. After processing using Independent Component Analysis (ICA), the signal was fed to an autoencoder for classification. By training the model properly we got a mean accuracy of 82.21% which is far better than the models using conventional machine learning methods. Our model achieved the state of the art results for classification of familiar and unfamiliar EEG signals.