Two SVDs produce more focal deep learning representations

Re-assign community