Convex Multiview Fisher Discriminant Analysis

CCA can be seen as a multiview extension of PCA, in which information from two sources is used for learning by finding a subspace in which the two views are most correlated. However PCA, and by extension CCA, does not use label information. Fisher Discriminant Analysis uses label information to find informative projections, which can be more informative in supervised learning settings. We derive a disciplined convex multi-view equivalent of kernel Fisher Discriminant Analysis. We then extend the optimisation problem to account for directions unique to each view as well the prominent direction across all views. We show experimental results on an EEG dataset and part of the PASCAL 2007 VOC challenge dataset.