Bayesian Approaches for adaptive Brain-Reading

It is to research how prior information about a problem can be incorporated into a classifier. The application scenario for this are Brain Computer Interfaces. Before every use of a BCI, a classifier has to be trained with previously recorded training data. It has been shown that data or classifier form other subjects can be used to shorten or totally skip the training. In this thesis the prior information is incorporated into the Bayesian Linear Discriminant Analysis to obtain high performance and in session adaptation.