Low-Rank Compression of Neural Nets: Learning the Rank of Each Layer | IEEE Conference Publication | IEEE Xplore