Development and Validation of a System for Classifying Imaginary Movements through a Brain-Computer Interface
Date:
MSc Dissertation (in Italian).
This thesis studies the development of algorithms for feature generation, feature selection, feature projection and classification for a non-invasive EEG-based sensory-motor brain-computer interface. Several algorithms were coded in Matlab; their performances were analysed, improved and statistically validated. An original genetic algorithm for feature generation that we crafted for this task provided the best performances when compared with traditional machine learning algorithms.