From learning metrics towards dependency exploration
We have recently introduced new kinds of data fusion techniques, where the goal is to find what is shared by data sets, instead of modeling all variation in data. They extend our earlier works on learning of distance metrics, discriminative clustering, and other supervised statistical data mining methods. In the new methods the supervision is symmetric, which translates to mining of dependencies. We have so far introduced methods for associative clustering and for extracting dependent components which generalize classical canonical correlations.