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 Hervé Jégou

Emerging Properties in Self-Supervised Vision Transformers


May 24, 2021
Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, Armand Joulin

* 21 pages 

  Access Paper or Ask Questions

ResMLP: Feedforward networks for image classification with data-efficient training


May 07, 2021
Hugo Touvron, Piotr Bojanowski, Mathilde Caron, Matthieu Cord, Alaaeldin El-Nouby, Edouard Grave, Armand Joulin, Gabriel Synnaeve, Jakob Verbeek, Hervé Jégou


  Access Paper or Ask Questions

Gradient-based Adversarial Attacks against Text Transformers


Apr 15, 2021
Chuan Guo, Alexandre Sablayrolles, Hervé Jégou, Douwe Kiela


  Access Paper or Ask Questions

Going deeper with Image Transformers


Apr 07, 2021
Hugo Touvron, Matthieu Cord, Alexandre Sablayrolles, Gabriel Synnaeve, Hervé Jégou


  Access Paper or Ask Questions

LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference


Apr 02, 2021
Ben Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Hervé Jégou, Matthijs Douze


  Access Paper or Ask Questions

Training Vision Transformers for Image Retrieval


Feb 10, 2021
Alaaeldin El-Nouby, Natalia Neverova, Ivan Laptev, Hervé Jégou


  Access Paper or Ask Questions

Training data-efficient image transformers & distillation through attention


Jan 15, 2021
Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablayrolles, Hervé Jégou


  Access Paper or Ask Questions

Grafit: Learning fine-grained image representations with coarse labels


Nov 25, 2020
Hugo Touvron, Alexandre Sablayrolles, Matthijs Douze, Matthieu Cord, Hervé Jégou


  Access Paper or Ask Questions

Powers of layers for image-to-image translation


Aug 13, 2020
Hugo Touvron, Matthijs Douze, Matthieu Cord, Hervé Jégou


  Access Paper or Ask Questions

Fixing the train-test resolution discrepancy: FixEfficientNet


Apr 20, 2020
Hugo Touvron, Andrea Vedaldi, Matthijs Douze, Hervé Jégou


  Access Paper or Ask Questions

Radioactive data: tracing through training


Feb 03, 2020
Alexandre Sablayrolles, Matthijs Douze, Cordelia Schmid, Hervé Jégou


  Access Paper or Ask Questions

White-box vs Black-box: Bayes Optimal Strategies for Membership Inference


Aug 29, 2019
Alexandre Sablayrolles, Matthijs Douze, Yann Ollivier, Cordelia Schmid, Hervé Jégou


  Access Paper or Ask Questions

And the Bit Goes Down: Revisiting the Quantization of Neural Networks


Jul 29, 2019
Pierre Stock, Armand Joulin, Rémi Gribonval, Benjamin Graham, Hervé Jégou


  Access Paper or Ask Questions

Fixing the train-test resolution discrepancy


Jul 19, 2019
Hugo Touvron, Andrea Vedaldi, Matthijs Douze, Hervé Jégou


  Access Paper or Ask Questions

Large Memory Layers with Product Keys


Jul 10, 2019
Guillaume Lample, Alexandre Sablayrolles, Marc'Aurelio Ranzato, Ludovic Denoyer, Hervé Jégou


  Access Paper or Ask Questions

Billion-scale semi-supervised learning for image classification


May 02, 2019
I. Zeki Yalniz, Hervé Jégou, Kan Chen, Manohar Paluri, Dhruv Mahajan


  Access Paper or Ask Questions

MultiGrain: a unified image embedding for classes and instances


Apr 03, 2019
Maxim Berman, Hervé Jégou, Andrea Vedaldi, Iasonas Kokkinos, Matthijs Douze


  Access Paper or Ask Questions

Equi-normalization of Neural Networks


Feb 27, 2019
Pierre Stock, Benjamin Graham, Rémi Gribonval, Hervé Jégou

* ICLR 2019 camera-ready 

  Access Paper or Ask Questions

Understanding and Improving Kernel Local Descriptors


Nov 27, 2018
Arun Mukundan, Giorgos Tolias, Andrei Bursuc, Hervé Jégou, Ondřej Chum


  Access Paper or Ask Questions

Déjà Vu: an empirical evaluation of the memorization properties of ConvNets


Sep 17, 2018
Alexandre Sablayrolles, Matthijs Douze, Cordelia Schmid, Hervé Jégou


  Access Paper or Ask Questions

Low-shot learning with large-scale diffusion


Jun 15, 2018
Matthijs Douze, Arthur Szlam, Bharath Hariharan, Hervé Jégou


  Access Paper or Ask Questions