Feature Reduction Using Genetic Algorithm for Cognitive Man-Machine Communication

Feature Reduction Using Genetic Algorithm for Cognitive Man-Machine Communication

Naveen Irtiza (Electrical Engineering Department, Bahria University, Karachi, Pakistan) and Humera Farooq (Computer Science Department, Bahria University, Karachi, Pakistan)
DOI: 10.4018/IJSSCI.2015100101
OnDemand PDF Download:
No Current Special Offers


Electroencephalographic (EEG) signals are usually comprised of high-dimensional feature space. This work aims to assess the effect of reducing the number of features extracted from EEG recordings. A methodology is proposed that combines brain imaging and machine learning techniques to predict the cognitive state of the subjects whether they are feeling themselves in a safe or dangerous environment. The changes in the brain state are correlated with power modulations of oscillatory rhythms in recorded EEG signals called ERD / ERS (Event-related De-synchronization / Synchronization). In order to determine the optimized number of features, Genetic Algorithm (GA) will be used. GA has played instrumental role in solving optimization problems from diverse fields. In various studies and researches for Cognitive Man-Machine Communication, the algorithm has been used as an effective method to extract an optimal set of features.
Article Preview

1. Introduction

A brain-computer interface (BCI) or a Cognitive man-machine communication interface can be defined as a communication system used to translate the brainwaves of a user into commands interpretable by a computer, bypassing the usual muscular channels. It enables movement-free communication where an EEG signal modulated by the user performing a particular cognitive task e.g., imagined movement or observing visually or auditory stimuli can be used to determine the user’s intent.

The most common method towards BCI is to analyze, categorize and interpret electroencephalographic signals (EEG), in such a way that they alter the state of a machine. The electrical activity of the brain is comprised of six different oscillatory rhythms characterized by their frequency bands. The rhythms are divided on the basis of their frequency ranges and location of origin namely delta (less than 4 Hz), theta (4-7 Hz), alpha (8-15 Hz), beta (12- 30 Hz), mu (8-12Hz) and gamma (32+Hz). Numerous existing studies have been performed to explore and investigate the inter-relationship between behavioral decisions and choices and oscillatory dynamics that accompany them (Dähne et al., 2014; Jatupaiboon, Pan-ngum, & Israsena, 2013).

On the other hand, several existing studies have been reported working on BCI for bio medical applications (Lantz, Grave de Peralta, Spinelli, Seeck, & Michel, 2003; Pfurtscheller, Müller-Putz, Pfurtscheller, & Rupp, 2005).In an application, a subject with complete paralysis of his left hand is equipped with an FES (Functional Electrical Stimulation) system. The system uses electrodes placed on the forearm of the subject, in order to send him an electrical current which forces his muscles to get tense, a task that the subject is not able to perform voluntarily (Pfurtscheller et al., 2005).

The Berlin group in Germany has worked on the development of a BCI based application known as Hex-O-Spell or the P300 speller which is a brain actuated spelling application (Blankertz et al., 2007). In this application, the subject has to control the movement and size of the arrow displayed on the screen using motor imagery to select a cell in a Hexagon consisting of 6 cells where each cell contains a group of letters or a letter.

In order to allow EEGs performed in one laboratory be reproduced in another, the 10-20 system, an international system of electrode placement, was introduced during the 1950s (Herwig, Satrapi, & Schönfeldt-Lecuona, 2003). This system uses several distinctive landmarks to help researchers record EEG signals related to the tasks of interest. Figure 1 shows a top view.

Figure 1.

The 10-20 system for placement of scalp electrodes (Jatupaiboon et al., 2013)


In the proceeding part of this section, various terminologies and concepts associated with EEG based BCIs are discussed. The next section 2 focuses on the problem statement. Along with this, the aim of the proposed work will also be stated. Section 3 describes the proposed methodology and Section 4 covers the conclusion and future directions of the proposed approach.

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 14: 4 Issues (2022): Forthcoming, Available for Pre-Order
Volume 13: 4 Issues (2021): 2 Released, 2 Forthcoming
Volume 12: 4 Issues (2020)
Volume 11: 4 Issues (2019)
Volume 10: 4 Issues (2018)
Volume 9: 4 Issues (2017)
Volume 8: 4 Issues (2016)
Volume 7: 4 Issues (2015)
Volume 6: 4 Issues (2014)
Volume 5: 4 Issues (2013)
Volume 4: 4 Issues (2012)
Volume 3: 4 Issues (2011)
Volume 2: 4 Issues (2010)
Volume 1: 4 Issues (2009)
View Complete Journal Contents Listing