Representation Learning Project
NOTE: a new theano version that we have tested to be correct will be published soon! Hold tight
This library contains variants of canonical correlation analysis: Kernal CCA, splitAE, and Deep CCA. The data used was the Wisconsin X-Ray Microbeam speech data set (XRMB).
Look at files in src/ directory. Particularly cca.m, kcca.m, dcca.py, dpca.py, and the ipython notebook files (for splitAE). Supporting files such as isvd.m (courtesy of Dr. Arora) scalableKCCA.m, and anything with a .mat extension are also in the directory but can be ignored.
Other standalone software packages were tested, they didn't work, thus they are in the IGNORE directory.
Please see the paper in the finalReport directory.