ABSTRACT
This paper intends to design a very simplified model of the human brain that learns tasks by mimicking the behavior of brain patterns. An EEG (Electroencephalography) signal is a physiological approach to record the electrical signal pertaining to the brain activity by placing the iron affinity electrode i.e. metal disc sensors on the scalp. The magnitude of this signal is quite small, measured in microvolts. There are five types of waves, namely-Alpha, Beta, Theta, Gamma and Delta. This EEG signal is acquired by placing the electrode at the frontal and temporal lobe positions which are then sampled and amplified using Original Muse. MATLAB or Python (open-source libraries) will be used for the processing of the raw EEG signal. This raw EEG signal denoting a band of thought impulse is fed to deep learning framework by threshold sampling. The network will be trained by supervised learning algorithm and simultaneously made to learn robust high-level feature from the processed EEG signal. Further testing of the neural network with a few predefined words under numerous test conditions will be done.
