Classification of Frontal EEG Signals of Normal Subjects to Differentiate Gender by Using Artificial Neural Network

Abstract

Varying mental states of an individual can influence their brainwave patterns. This is also true for individuals of different gender, where a male’s EEG signal is different from a female’s EEG signal. This provides a context for our research, where our main aim is to classify different patterns of EEG based on different gender. This paper presents our initial study to classify gender of normal subjects based on their frontal EEG signals. Forty normal subjects have participated in this experiment, where their EEG signals have been recorded for analysis purpose. The recorded raw EEG data is first pre-processed and filtered into 4 different frequency subbands. Two types of analysis were then conducted; the first analysis took into consideration all four sub-bands of frontal EEG, whereas the second analysis only considered two subbands namely alpha and beta bands. The features extracted from the selected sub-bands are in the form of EEG Energy Spectral Density (ESD) values, which are then fed into an Artificial Neural Network (ANN) classifier for classification purpose; i.e. to distinguish between male and female. Based on results obtained from the analysis, it is found that higher classification accuracy can be achieved from combining four sub-bands when compared to if only two sub-bands (alpha and beta) are being considered.