Brain-computer interface is a promising research area that has the potential to aid
impaired individuals in their daily lives. There are several different methods for
capturing brain signals, both invasive and noninvasive. A popular noninvasive
technique is electroencephalography (EEG). It is of great interest to be able to
interpret EEG signals accurately so that a machine can carry out correct instructions.
This paper looks at different machine learning techniques, both linear and nonlinear,
in an attempt to classify EEG signals. It is found that support vector machines provide
more satisfactory results than neural networks.