Electroencephalogram (EEG) signal processing, wave identification, and emotion recognition

One of the missing links between man and machine today is the ability for a machine to recognize and act on human emotion. Now, with human computer interfacing hardware, such as electroencephalogram (EEG) neuroheadsets, becoming more readily available to the commercial market, the dynamic map of the human brain can be integrated into our increasingly digital lifestyles. This project will begin with first acquiring raw EEG data, then perform various digital signal processing algorithms to extract the pertinent data that lies within a raw recording of an EEG signal, use the method of auto-regression modeling to model an estimated power spectral density to assist in identifying dominant wave patterns, calculate power magnitude contribution per wave band (i.e. delta, theta, alpha, beta, gamma bands), use the method of principle component analysis to reduce the dimensionality of the large EEG datasets, then finally use various methods of kernel functions and hyperplane solution methods to construct support vector machine models to correctly classify'positive' and'negative' human emotion from EEG patterns with the highest fidelity possible.