activity
20172022
most citedHigh-Performance Large-Scale Image Recognition Without Normalization

256 citations · 436 across the 6 of their papers we have counts for

collaborators
Showing cs.LGShow all

8 papers · 1 filter

cs.LG20223 cited

Regularising for invariance to data augmentation improves supervised learning

Aleksander Botev, Matthias Bauer, Soham De

Data augmentation is used in machine learning to make the classifier invariant to label-preserving transformations. Usually this invariance is only encouraged implicitly by includi…

cs.LG202140 cited

On the Origin of Implicit Regularization in Stochastic Gradient Descent

Samuel L. Smith, Benoit Dherin, David G. T. Barrett +1

For infinitesimal learning rates, stochastic gradient descent (SGD) follows the path of gradient flow on the full batch loss function. However moderately large learning rates can a…

cs.LG202121 cited

Characterizing signal propagation to close the performance gap in unnormalized ResNets

Andrew Brock, Soham De, Samuel L. Smith

Batch Normalization is a key component in almost all state-of-the-art image classifiers, but it also introduces practical challenges: it breaks the independence between training ex…

cs.LG202022 cited

On the Generalization Benefit of Noise in Stochastic Gradient Descent

Samuel L. Smith, Erich Elsen, Soham De

It has long been argued that minibatch stochastic gradient descent can generalize better than large batch gradient descent in deep neural networks. However recent papers have quest…

cs.LG2020

Batch Normalization Biases Residual Blocks Towards the Identity Function in Deep Networks

Soham De, Samuel L. Smith

Batch normalization dramatically increases the largest trainable depth of residual networks, and this benefit has been crucial to the empirical success of deep residual networks on…

cs.LG2019

The Impact of Neural Network Overparameterization on Gradient Confusion and Stochastic Gradient Descent

Karthik A. Sankararaman, Soham De, Zheng Xu +2

This paper studies how neural network architecture affects the speed of training. We introduce a simple concept called gradient confusion to help formally analyze this. When gradie…