papers AI Learner
The Github is limit! Click to go to the new site.

Equi-normalization of Neural Networks

2019-02-27
Pierre Stock, Benjamin Graham, Rémi Gribonval, Hervé Jégou

Abstract

Modern neural networks are over-parametrized. In particular, each rectified linear hidden unit can be modified by a multiplicative factor by adjusting input and output weights, without changing the rest of the network. Inspired by the Sinkhorn-Knopp algorithm, we introduce a fast iterative method for minimizing the L2 norm of the weights, equivalently the weight decay regularizer. It provably converges to a unique solution. Interleaving our algorithm with SGD during training improves the test accuracy. For small batches, our approach offers an alternative to batch-and group-normalization on CIFAR-10 and ImageNet with a ResNet-18.

Abstract (translated by Google)
URL

http://arxiv.org/abs/1902.10416

PDF

http://arxiv.org/pdf/1902.10416


Similar Posts

Comments