Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox

Picture for Samuel L. Smith

Drawing Multiple Augmentation Samples Per Image During Training Efficiently Decreases Test Error


May 27, 2021
Stanislav Fort, Andrew Brock, Razvan Pascanu, Soham De, Samuel L. Smith


  Access Paper or Ask Questions

High-Performance Large-Scale Image Recognition Without Normalization


Feb 11, 2021
Andrew Brock, Soham De, Samuel L. Smith, Karen Simonyan


  Access Paper or Ask Questions

On the Origin of Implicit Regularization in Stochastic Gradient Descent


Jan 28, 2021
Samuel L. Smith, Benoit Dherin, David G. T. Barrett, Soham De

* Accepted as a conference paper at ICLR 2021 

  Access Paper or Ask Questions

Characterizing signal propagation to close the performance gap in unnormalized ResNets


Jan 27, 2021
Andrew Brock, Soham De, Samuel L. Smith

* Published as a conference paper at ICLR 2021 

  Access Paper or Ask Questions

Cold Posteriors and Aleatoric Uncertainty


Jul 31, 2020
Ben Adlam, Jasper Snoek, Samuel L. Smith

* ICML workshop on Uncertainty and Robustness in Deep Learning (2020) 
* 5 pages, 3 figures 

  Access Paper or Ask Questions

On the Generalization Benefit of Noise in Stochastic Gradient Descent


Jun 26, 2020
Samuel L. Smith, Erich Elsen, Soham De

* Camera-ready version of ICML 2020 

  Access Paper or Ask Questions

Batch Normalization Biases Deep Residual Networks Towards Shallow Paths


Feb 24, 2020
Soham De, Samuel L. Smith


  Access Paper or Ask Questions

The Effect of Network Width on Stochastic Gradient Descent and Generalization: an Empirical Study


May 09, 2019
Daniel S. Park, Jascha Sohl-Dickstein, Quoc V. Le, Samuel L. Smith

* 17 pages, 3 tables, 17 figures; accepted to ICML 2019 

  Access Paper or Ask Questions

Stochastic natural gradient descent draws posterior samples in function space


Oct 16, 2018
Samuel L. Smith, Daniel Duckworth, Semon Rezchikov, Quoc V. Le, Jascha Sohl-Dickstein

* 11 pages, 6 figures 

  Access Paper or Ask Questions

Decoding Decoders: Finding Optimal Representation Spaces for Unsupervised Similarity Tasks


May 09, 2018
Vitalii Zhelezniak, Dan Busbridge, April Shen, Samuel L. Smith, Nils Y. Hammerla

* ICLR 2018 Workshop Track, 15 pages, 3 figures, 6 tables 

  Access Paper or Ask Questions

Don't Decay the Learning Rate, Increase the Batch Size


Feb 24, 2018
Samuel L. Smith, Pieter-Jan Kindermans, Chris Ying, Quoc V. Le

* 11 pages, 8 figures. Published as a conference paper at ICLR 2018 

  Access Paper or Ask Questions

A Bayesian Perspective on Generalization and Stochastic Gradient Descent


Feb 14, 2018
Samuel L. Smith, Quoc V. Le

* 13 pages, 9 figures. Published as a conference paper at ICLR 2018 

  Access Paper or Ask Questions

Offline bilingual word vectors, orthogonal transformations and the inverted softmax


Feb 13, 2017
Samuel L. Smith, David H. P. Turban, Steven Hamblin, Nils Y. Hammerla

* Accepted to conference track at ICLR 2017 

  Access Paper or Ask Questions