EPFL Papers @ NeurIPS 2020

33rd NeurIPS conference in Vancouver, Canada © Khari Johnson / VentureBeat

33rd NeurIPS conference in Vancouver, Canada © Khari Johnson / VentureBeat

A list of accepted papers for NeurIPS 2020 has been announced and EPFL scores an impressive number of 41 accepted papers.
Due to Covid-19, this year NeurIPS edition will be a virtual-only conference held from Sunday, December 6th through Saturday, December 12th.
The list of accepted papers from EPFL is as follows:

DISK: Learning local features with policy gradient
Michał Tyszkiewicz (EPFL) · Pascal Fua (EPFL, Switzerland) · Eduard Trulls (Google)

MeshSDF: Differentiable Iso-Surface Extraction
Edoardo Remelli (EPFL) · Artem Lukoyanov (EPFL) · Stephan Richter (Intel Labs) · Benoit Guillard (EPFL) · Timur Bagautdinov (Facebook) · Pierre Baque (Neural Concept SA) · Pascal Fua (EPFL, Switzerland)

Minimax classification with 0-1 loss and performance guarantees
Santiago Mazuelas (Basque Center for Applied Mathematics) · Andrea Zanoni (EPFL) · Aritz Pérez (Basque Center for Applied Mathematics (BCAM))

ExpandNets: Linear Over-parameterization to Train Compact Convolutional Networks
Shuxuan Guo (EPFL) · Jose M. Alvarez (NVIDIA) · Mathieu Salzmann (EPFL)

Complex Dynamics in Simple Neural Networks: Understanding Gradient Flow in Phase Retrieval
Stefano Sarao Mannelli (Institut de Physique Théorique) · Giulio Biroli (ENS) · Chiara Cammarota (King's College London) · Florent Krzakala (ENS Paris & Sorbonnes Université) · Pierfrancesco Urbani (Institut de Physique Théorique) · Lenka Zdeborová (CEA Saclay & EPFL)

Neural Anisotropy Directions
Guillermo Ortiz-Jimenez (EPFL) · Apostolos Modas (EPFL) · Seyed-Mohsen Moosavi-Dezfooli (ETHZ) · Pascal Frossard (EPFL)

Hold me tight! Influence of discriminative features on deep network boundaries
Guillermo Ortiz-Jimenez (EPFL) · Apostolos Modas (EPFL) · Seyed-Mohsen Moosavi-Dezfooli (ETHZ) · Pascal Frossard (EPFL)

Fairness in Streaming Submodular Maximization: Algorithms and Hardness
Marwa El Halabi (MIT) · Slobodan Mitrović (MIT) · Ashkan Norouzi-Fard (Google Research) · Jakab Tardos (EPFL) · Jakub Tarnawski (Microsoft Research)

Practical Low-Rank Communication Compression in Decentralized Deep Learning
Thijs Vogels (EPFL) · Sai Praneeth Karimireddy (EPFL) · Martin Jaggi (EPFL)

Continual Deep Learning by Functional Regularisation of Memorable Past
Pingbo Pan (University of Technology Sydney) · Siddharth Swaroop (University of Cambridge) · Alexander Immer (EPFL) · Runa Eschenhagen (University of Osnabrueck) · Richard E Turner (University of Cambridge) · Mohammad Emtiyaz Khan (RIKEN, Tokyo)

How hard is to distinguish graphs with graph neural networks?
Andreas Loukas (EPFL)

Optimization and Generalization of Shallow Neural Networks with Quadratic Activation Functions
Stefano Sarao Mannelli (Institut de Physique Théorique) · Eric Vanden-Eijnden (New York University) · Lenka Zdeborová (CEA Saclay & EPFL)

Generalization error in high-dimensional perceptrons: Approaching Bayes error with convex optimization
Benjamin Aubin (Ipht Saclay) · Florent Krzakala (ENS Paris & Sorbonnes Université) · Yue Lu (Harvard University) · Lenka Zdeborová (CEA Saclay & EPFL)

Building powerful and equivariant graph neural networks with message-passing
Clément Vignac (EPFL) · Andreas Loukas (EPFL) · Pascal Frossard (EPFL)

Online Robust Regression via SGD on the l1 loss
Scott Pesme (EPFL) · Nicolas Flammarion (EPFL)

The Devil is in the Detail: a Framework for Macroscopic Prediction via Microscopic Models
Yingxiang Yang (ByteDance) · Negar Kiyavash (EPFL) · Le Song (Georgia Institute of Technology) · Niao He (UIUC)

Information theoretic limits of learning a sparse rule
Clément Luneau (EPFL) · Jean Barbier (EPFL) · Nicolas Macris (EPFL)

On the Loss Landscape of Adversarial Training: Identifying Challenges and How to Overcome Them
Chen Liu (EPFL) · Mathieu Salzmann (EPFL) · Tao LIN (EPFL) · Ryota Tomioka (Microsoft Research Cambridge) · Sabine Süsstrunk (EPFL)

WoodFisher: Efficient Second-Order Approximation for Neural Network Compression
Sidak Pal Singh (EPFL) · Dan Alistarh (IST Austria & Neural Magic Inc.)

Adaptive Gradient Methods Converge Under Heavy-tailed Noise
Jingzhao Zhang (MIT) · Sai Praneeth Karimireddy (EPFL) · Andreas Veit (Google) · Seungyeon Kim (Google Research) · Sashank Reddi (Google) · Sanjiv Kumar (Google Research) · Suvrit Sra (MIT)

Learning Augmented Energy Minimization via Speed Scaling
Etienne Bamas (EPFL) · Andreas Maggiori (EPFL) · Lars Rohwedder (EPFL) · Ola Svensson (EPFL)

A Catalyst Framework for Minimax Optimization
Junchi Yang (University of Illinois) · Siqi Zhang (University of Illinois at Urbana-Champaign) · Negar Kiyavash (EPFL) · Niao He (UIUC)

DeepSVG: A Hierarchical Generative Network for Vector Graphics Animation
Alexandre Carlier (ETH Zurich) · Martin Danelljan (ETH Zurich) · Alexandre Alahi (EPFL) · Radu Timofte (ETH Zurich)

The Primal-Dual method for Learning Augmented Algorithms
Etienne Bamas (EPFL) · Andreas Maggiori (EPFL) · Ola Svensson (EPFL)

Reservoir Computing meets Recurrent Kernels and Structured Transforms
Jonathan Dong (Laboratoire Kastler-Brossel) · Ruben Ohana (Ecole Normale Supérieure) · Mushegh Rafayelyan (Kastler-Brossel Laboratory (ENS, Sorbonne U., PSL U., CNRS, Collège de France)) · Florent Krzakala (ENS Paris & Sorbonnes Université & EPFL)

Global Convergence and Variance Reduction for a Class of Nonconvex-Nonconcave Minimax Problems
Junchi Yang (University of Illinois) · Negar Kiyavash (EPFL) · Niao He (UIUC)

On the Almost Sure Convergence of Stochastic Gradient Descent in Non-Convex Problems
Panayotis Mertikopoulos (CNRS (French National Center for Scientific Research)) · Nadav Hallak (EPFL) · Ali Kavis (EPFL) · Volkan Cevher (EPFL)

Direct Feedback Alignment Scales to Modern Deep Learning Tasks and Architectures
Julien Launay (LightOn) · Iacopo Poli (LightOn) · François Boniface (LightOn) · Florent Krzakala (ENS Paris & Sorbonnes Université & EPFL)

Kernel Alignment Risk Estimator: Risk Prediction from Training Data
Arthur Jacot (EPFL) · Berfin Simsek (EPFL) · Francesco Spadaro (EPFL) · Clement Hongler (EPFL) · Franck Gabriel (EPFL)

Robust Reinforcement Learning via Adversarial training with Langevin Dynamics
Parameswaran Kamalaruban (EPFL) · Yu-Ting Huang (EPFL) · Ya-Ping Hsieh (EPFL) · Paul Rolland (EPFL) · Cheng Shi (Unversity of Basel) · Volkan Cevher (EPFL)

Understanding and Improving Fast Adversarial Training
Maksym Andriushchenko (EPFL) · Nicolas Flammarion (EPFL)

Erdos Goes Neural: an Unsupervised Learning Framework for Combinatorial Optimization on Graphs
Nikolaos Karalias (EPFL) · Andreas Loukas (EPFL)

Fast and Accurate k-means++ via Rejection Sampling
Vincent Cohen-Addad (CNRS & Sorbonne Université) · Silvio Lattanzi (Google Research) · Ashkan Norouzi-Fard (Google Research) · Christian Sohler (University of Cologne) · Ola Svensson (EPFL)

Phase retrieval in high dimensions: Statistical and computational phase transitions
Antoine Maillard (Ecole Normale Supérieure) · Bruno Loureiro (IPhT Saclay) · Florent Krzakala (ENS Paris & Sorbonnes Université) · Lenka Zdeborová (CEA Saclay & EPFL)

Dynamical mean-field theory for stochastic gradient descent in Gaussian mixture classification
Francesca Mignacco (IPhT, CEA Saclay) · Florent Krzakala (ENS Paris & Sorbonnes Université) · Pierfrancesco Urbani (Institut de Physique Théorique) · Lenka Zdeborová (CEA Saclay & EPFL)

Coresets for Regressions with Panel Data
Lingxiao Huang (EPFL) · K Sudhir (Yale University) · Nisheeth Vishnoi (Yale University)

Model Fusion via Optimal Transport
Sidak Pal Singh (EPFL) · Martin Jaggi (EPFL)

Ensemble Distillation for Robust Model Fusion in Federated Learning
Tao Lin (EPFL) · Lingjing Kong (EPFL) · Sebastian U Stich (EPFL) · Martin Jaggi (EPFL)

UCLID-Net: Single View Reconstruction in Object Space
Benoit Guillard (EPFL) · Edoardo Remelli (EPFL) · Pascal Fua (EPFL, Switzerland)

All-or-nothing statistical and computational phase transitions in sparse spiked matrix estimation
Jean Barbier (EPFL) · Nicolas Macris (EPFL) · Cynthia Rush (Columbia University)

Deep Smoothing of the Implied Volatility Surface
Damien Ackerer (Swissquote) · Natasa Tagasovska (EPFL) · Thibault Vatter (Columbia University)