What if Neural Networks had SVDs?
Published:
Various Neural Networks employ time-consuming matrix operations like matrix inversion. Many such matrix operations are faster to compute given the Singular Value Decomposition (SVD). Techniques from [1, 2] allow using the SVD in Neural Networks without computing it. In theory, the techniques can speed up matrix operations, however, in practice, they are not fast enough.
We present an algorithm which is up to $27 \times $ faster than a previous approach, fast enough to speed up several matrix operations.