Poster
Preconditioned Spectral Descent for Deep Learning
David Carlson · Edo Collins · Ya-Ping Hsieh · Lawrence Carin · Volkan Cevher

Tue Dec 8th 07:00 -- 11:59 PM @ 210 C #30 #None

Deep learning presents notorious computational challenges. These challenges include, but are not limited to, the non-convexity of learning objectives and estimating the quantities needed for optimization algorithms, such as gradients. While we do not address the non-convexity, we present an optimization solution that ex- ploits the so far unused “geometry” in the objective function in order to best make use of the estimated gradients. Previous work attempted similar goals with preconditioned methods in the Euclidean space, such as L-BFGS, RMSprop, and ADA-grad. In stark contrast, our approach combines a non-Euclidean gradient method with preconditioning. We provide evidence that this combination more accurately captures the geometry of the objective function compared to prior work. We theoretically formalize our arguments and derive novel preconditioned non-Euclidean algorithms. The results are promising in both computational time and quality when applied to Restricted Boltzmann Machines, Feedforward Neural Nets, and Convolutional Neural Nets.

Author Information

David Carlson (Duke University)
Edo Collins
Ya-Ping Hsieh (EPFL)
Lawrence Carin (Duke University)
Volkan Cevher (EPFL)

More from the Same Authors