Part of Advances in Neural Information Processing Systems 33 (NeurIPS 2020)

*Xiaoxia Wu, Edgar Dobriban, Tongzheng Ren, Shanshan Wu, Zhiyuan Li, Suriya Gunasekar, Rachel Ward, Qiang Liu*

Normalization methods such as batch, weight, instance, and layer normalization are commonly used in modern machine learning. Here, we study the weight normalization (WN) method \cite{salimans2016weight} and a variant called reparametrized projected gradient descent (rPGD) for overparametrized least squares regression and some more general loss functions. WN and rPGD reparametrize the weights with a scale $g$ and a unit vector such that the objective function becomes \emph{non-convex}. We show that this non-convex formulation has beneficial regularization effects compared to gradient descent on the original objective. These methods adaptively regularize the weights and \emph{converge linearly} close to the minimum $\ell_2$ norm solution even for initializations far from zero. For certain two-phase variants, they can converge to the min norm solution. This is different from the behavior of gradient descent, which only converges to the min norm solution when started at zero, and thus more sensitive to initialization.

Do not remove: This comment is monitored to verify that the site is working properly