Practical Riemannian Neural Networks

Abstract

We provide the first experimental results on non-synthetic datasets for the quasi-diagonal Riemannian gradient descents for neural networks introduced in [Ollivier, 2015]. These include the MNIST, SVHN, and FACE datasets as well as a previously unpublished electroencephalogram dataset. The quasi-diagonal Riemannian algorithms consistently beat simple stochastic gradient gradient descents by a varying margin. The computational overhead with respect to simple backpropagation is around a factor $2$. Perhaps more interestingly, these methods also reach their final performance quickly, thus requiring fewer training epochs and a smaller total computation time. We also present an implementation guide to these Riemannian gradient descents for neural networks, showing how the quasi-diagonal versions can be implemented with minimal effort on top of existing routines which compute gradients.

Cited by in corpus (7)

• Fisher Information and Natural Gradient Learning of Random Deep Networks
• The Extended Kalman Filter is a Natural Gradient Descent in Trajectory Space
• When Does Preconditioning Help or Hurt Generalization?
• True Asymptotic Natural Gradient Optimization
• Diagonal Rescaling For Neural Networks
• Train Feedfoward Neural Network with Layer-wise Adaptive Rate via Approximating Back-matching Propagation
• Channel-Directed Gradients for Optimization of Convolutional Neural Networks