Human emotions classification using bag-of-words method on single electrode brain computer interface

Abstract

In this paper we present a human emotions classification technique based on EEG signals from single electrode. We designed an EEG experiment in which we collected training data. Applying the Daubechies 8 wavelet (db8) delta, theta, alpha, beta and gamma wave signal are obtained by decomposition and used as codewords for Bag-of-Words model. We represented each signal with its codewords histogram and implemented classifier base on minimum distance. Results show that EEG signals induced watching horror and relaxing video clips can be classified with average classification rate of 75% using minimum distance classifier.