Abstract:
Brain Computer Interface (BCI) systems have ushered a new era of neural engineering research. At the core of BCI research is development of data acquisition, filtration and classification techniques that can accurately decode the brain activity that occurs while performing a motor task. In this study we investigate the classification accuracy of Linear Discriminant Analysis, Quadratic Discriminant Analysis, Naïve Bayes, Quadratic Support Vector Machine and Radial Based Function Support Vector Machine and Multilayer perceptron classifiers for classifying the flexion/extension of forearm and wrist. Moreover, Hjorth Parameters and Power Spectral Density are employed as feature extraction techniques to derive four different feature vectors that are later used to train our classifiers. At the culmination of this study, it is shown that QDA classifier trained with average band power (PSD) feature vector has the highest classification accuracy at 80.20% followed by Quadratic Support Vector Machine trained with Activity feature vector at 76.92%. Apart from enhancing accuracy of classifying the four fundamental upper limb movements, this study will eventually contribute towards developing better controllers for neuro-prosthetic devices. The study has been performed experimentally with Emotiv headsets equipped with fourteen electrodes to acquire EEG data from two human test subjects in synchronous mode. Classification and data analysis has been performed offline however in future the study will be extended to an online BCI system.
