Generalized eigendecomposition is a powerful method of spatial filtering in order to extract components from the data. You'll learn the theory, motivations, and see a few examples. Also discussed is the dangers of overfitting noise and few ways to avoid it. It is about 50 minutes long (sorry, a bit on the long side).
|This zip file contains all of the data and scripts for this lecturelet. It is 300 MB. You'll also need the eeglab toolbox to use the topoplot function (or comment those lines out of the code).|