Abstract:
The sever accidents, paralysis attacks and different diseases are major cause of interrupting
the normal communication channel of brain with other body parts. The individuals, victim of
above sever cases can’t live a normal life and are burden on the society. To help these
affected people who may have partially or completely lost independent motions of their
limbs, there is requirement of a system which can bypass the normal communication channel
of the brain and sends messages to the exterior world. To implement such kind of system,
acquiring, filtering, feature extraction and classifying the brain signals is a major task. The
focus of this research is to classify the EEG signals dataset by Artificial Neural Network,
Support Vector Machine and well-known statistical techniques e.g. Linear Discriminant
Analysis, Quadratic Discriminant Analysis, Naive Bayes and Decision Trees and also
compare them to identify a suitable technique for hardware implementation. The
performances of these classifiers are compared on the basis of confusion matrix and mean
square error. The most efficient method will be used to give signals to microcontroller to
control the motion of 2-DoF Robotic Manipulator for Upper Limb Prosthesis. The 2-DoF
Robotic Manipulator designed in NUST will be used for testing. This research enhances the
future potential capabilities of BCI systems.
