This project belongs to the brain-computer interface (BCI) research field. According to the project proposal, we have developed a tool for aquiring and processing electroencefalographic (EEG) data in real time. It also learns patterns from the EEG data by means of Neural Networks and allows a person to use his thoughts to control a cursor on the screen.
We have also proposed and tested new ideas to improve the accuracy of EEG classification for BCIs. In particular, we have developped a method based on transition detection between thoughts. Transition detection improves accuracy by reducing the number of classes. The moving window also improves accuracy by computing the final classification from a set of predictions (the window), instead of a single one.
The group has also started to apply evolutionary multiobjective machine learning techniques to EEG classification.