Alert button

Controlling Covariate Shift using Equilibrium Normalization of Weights

Add code
Bookmark button
Alert button
Dec 11, 2018
Aaron Defazio, Léon Bottou

Figure 1 for Controlling Covariate Shift using Equilibrium Normalization of Weights
Figure 2 for Controlling Covariate Shift using Equilibrium Normalization of Weights
Figure 3 for Controlling Covariate Shift using Equilibrium Normalization of Weights
Figure 4 for Controlling Covariate Shift using Equilibrium Normalization of Weights

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon

We introduce a new normalization technique that exhibits the fast convergence properties of batch normalization using a transformation of layer weights instead of layer outputs. The proposed technique keeps the contribution of positive and negative weights to the layer output in equilibrium. We validate our method on a set of standard benchmarks including CIFAR-10/100, SVHN and ILSVRC 2012 ImageNet.

View paper onarxiv iconopen_review iconOpenReview

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon