EPFL CIS NeurIPS 2021 Mirror Event

 A one-day mirror event on machine learning on December 6

The Conference on Neural Information Processing Systems (NeurIPS) is the main machine learning conference.

While the conference being highly selective, EPFL researchers have more than 60 papers accepted for presentation at the conference, more than any other continental Europe institutions. However, due to the current pandemic, the conference will be done entirely remotely instead of in-person.

We have thus decided to organize a local mid-scale event in SwissTech Convention Center, on the EPFL campus, and to invite all the EPFL researchers and students who have a paper accepted at NeurIPS this year to join us on Monday, December 6, for an in-person poster session, and extended the invitation to all Swiss institutions. We do believe that local events such as this one are also very useful to reduce our travel impact while creating opportunities for informal discussion and creation of new ideas.

The conference will give the occasion to all EPFL students and researchers with accepted papers at NeurIPS 2021 to present their works, and to all students and researchers interested in machine learning research to connect and discuss science during this one-day event.


Organisation committee:
Prof. Florent Krzakala, EPFL (IdePhics Lab.)
Prof. Nicolas Flammarion, EPFL (TML Lab.)
Prof. Lenka Zdeborova, EPFL (SPOC Lab.)

Program

09:00-10:00 Welcome coffee

10:00-11:00 Spotlight Talks: 9 x Spotlight session (5 mins each)

  • Decentralized Learning in Online Queuing Systems, Etienne Boursier
  • Learning Gaussian Mixtures with Generalized Linear Models: Precise Asymptotics in High-dimensions Bruno Loureiro
  • Uniform Sampling over Episode Difficulty, Séb Arnold
  • Sequential Algorithms for Testing Closeness of Distributions, Aadil Oufkir
  • Prototypical Cross-Attention Networks for Multiple Object Tracking and Segmentation, Lei Ke
  • Repulsive Deep Ensembles are Bayesian, Vincent Fortuin
  • On the Power of Differentiable Learning versus PAC and SQ Learning, Emmanuel Abbe
  • Credit Assignment in Neural Networks through Deep Feedback Control, Alexander Meulemans, Matilde Tristany Farinha
  • Precise characterization of the prior predictive
    distribution of deep ReLU networks, Lorenzo Noci, Gregor Bachmann

11:00-12:00 poster sessions (see listing below)

12h00-14:00 lunch on campus

14:00-15:00 Oral Session (25mins each)

  • Stability and Deviation Optimal Risk Bounds with Convergence Rate O(1/n), Nikita Zhivotovskiy
  • A Continuized View on Nesterov Acceleration for Stochastic Gradient Descent and Randomized Gossip, Raphael Berthier

15:00-17:00 Coffee & poster sessions

Poster list

1* RobustBench: a standardized adversarial robustness benchmark    Other NeurIPS tracks (e.g. competition, benchmark…)    Francesco Croce, Maksym Andriushchenko, Vikash Sehwag, Edoardo Debenedetti, Nicolas Flammarion, Mung Chiang, Prateek Mittal, Matthias Hein

2* Decentralized Learning in Online Queuing Systems    Spotlight    Flore Sentenac, Etienne Boursier, Vianney Perchet

3* Score-based Generative Neural Networks for Large-Scale Optimal Transport     Poster    Max Daniels, Tyler Maunu, Paul Hand

4* Generalization error rates in kernel regression : the crossover from the noiseless to the noisy regime    Poster    Hugo Cui, Bruno Loureiro, Florent Krzakala, Lenka Zdeborová

5*Learning curves of generic features maps for realistic datasets with a teacher-student model    Poster    Bruno Loureiro, Cedric Gerbelot, Hugo Cui, Sebastian Goldt, Florent Krzakala, Marc Mezard, Lenka Zdeborová

6* Learning Gaussian Mixtures with Generalized Linear Models: Precise Asymptotics in High-dimensions    Spotlight    Bruno Loureiro, Gabriele Sicuro, Cedric Gerbelot, Alessandro Pacco, Florent Krzakala, Lenka Zdeborova

7* What can linearized neural networks actually say about generalization?    Poster    Guillermo Ortiz-Jimenez, Seyed-Mohsen Moosavi-Dezfooli, Pascal Frossard

8*Last iterate convergence of SGD for Least-Squares in the Interpolation regime    Poster    Aditya Varre, Loucas Pillaud-Vivien, Nicolas Flammarion

9* Interpreting Language Models Through Knowledge Graph Extraction    Spotlight    Vinitra Swamy, Angelika Romanou, Martin Jaggi

10* Neural Design Space Exploration    Other NeurIPS tracks (e.g. competition, benchmark…)    Wei Jiang, Richard Lee Davis, Kevin Kim, Pierre Dillenbourg

11* Locality defeats the curse of dimensionality in convolutional teacher-student scenarios    Poster    A. Favero, F. Cagnetta, M. Wyart

12* An Improved Analysis of Gradient Tracking for Decentralized Machine Learning    Poster    Anastasia Koloskova, Tao Lin, Sebastian Stich

13* RelaySum for Decentralized Deep Learning on Heterogeneous Data    Poster    Thijs Vogels, Lie He, Anastasia Koloskova, Tao Lin, Sai Praneeth Karimireddy, Sebastian U. Stich, Martin Jaggi

14* Learning Transferable Adversarial Perturbations    Poster    Krishna Kanth Nakka and Mathieu Salzmann

15* Efficient and Local Parallel Random Walks    Poster    Michael Kapralov, Silvio Lattanzi, Navid Nouri, Jakab Tardos

16* Stability and Deviation Optimal Risk Bounds with Convergence Rate O(1/n)    Oral    Yegor Klochkov, Nikita Zhivotovskiy

17* Model, sample, and epoch-wise descents: exact solution of gradient flow in the random feature model    Poster    Antoine Bodin & Nicolas Macris

18* Relative stability toward diffeomorphisms indicates performance in deep nets    Poster    Leonardo Petrini, Alessandro Favero, Mario Geiger, Matthieu Wyart

19* TTT++: When Does Self-Supervised Test-Time Training Fail or Thrive?    Poster    Yuejiang Liu, Parth Kothari, Bastien Germain van Delft, Baptiste Bellot-Gurlet, Taylor Mordan, Alexandre Alahi

20* Implicit Bias of SGD for Diagonal Linear Networks: a Provable Benefit of Stochasticity.     Poster    Scott Pesme, Loucas Pillaud-Vivien, Nicolas Flammarion

21* A Continuized View on Nesterov Acceleration for Stochastic Gradient Descent and Randomized Gossip    Oral    M. Even, R. Berthier, F. Bach, N. Flammarion, P. Gaillard, H. Hendrikx, L. Massoulié, A. Taylor

22* Unfirom Sampling over Episode Difficulty    Spotlight    Sébastien M. R. Arnold, Guneet S. Dhillon, Avinash Ravichandran, Stefano Soatto

23* Distilling Image Classifiers in Object Detectors    Poster    Shuxuan Guo,  Jose M. Alvarez, Mathieu Salzmann

24* Robust Inverse Reinforcement Learning under Transition Dynamics Mismatch    Poster    Luca Viano, Yu-Ting Huang, Kamalaruban Parameswaran, Adrian Weller, Volkan Cevher

25* A first-order primal-dual method with adaptivity to local smoothness    Poster    Maria Vladarean, Yura Malitsky, Volkan Cevher

26* Collaborative Learning in the Jungle (Decentralized, Byzantine, Heterogeneous, Asynchronous and Nonconvex Learning)     Poster    El Mahdi El Mhamdi, Sadegh Farhadkhani, Rachid Guerraoui, Arsany Guirguis, Lê-Nguyên Hoang, Sébastien Rouault

27* Conditional Generation Using Polynomial Expansions    Poster    Grigorios Chrysos, Markos Georgopoulos, Yannis Panagakis

28* Sequential Algorithms for Testing Closeness of Distributions    Spotlight    Omar Fawzi, Nicolas Flammarion, Aurélien Garivier, Aadil Oufkir

29* Breaking the centralized barrier for cross-device federated learning    Poster    Sai Praneeth Karimireddy, Martin Jaggi, Satyen Kale, Mehryar Mohri, Sashank Reddi, Sebastian Stich, Ananda Theertha Suresh

30* SQALER: scaling question answering by decoupling multi-hop and logical reasoning    Poster    Mattia Atzeni, Jasmina Bogojeska, Andreas Loukas

31* Prototypical Cross-Attention Networks for Multiple Object Tracking and Segmentation    Spotlight    Lei Ke, Xia Li, Martin Danelljan, Yu-Wing Tai, Chi-Keung Tang, Fisher Yu

32* Making the most of your day: online learning for optimal allocation of time    Poster    Etienne Boursier, Tristan Garrec, Vianney Perchet, Marco Scarsini

33* Self-Supervised Representation Learning on Neural Network Weights for Model Characteristic Prediction    Poster    Konstantin Schürholt, Dimche Kostadinov, Damian Borth

34* On the Bias-Variance-Cost Tradeoff of Stochastic Optimization    Poster    Yifan Hu, Xin Chen, Niao He

35* Local plasticity rules can learn deep representations using self-supervised contrastive predictions    Poster    Bernd Illing, Jean Ventura, Guillaume Bellec, Wulfram Gerstner

36* Partition and Code: learning how to compress graphs    Poster    Giorgos Bouritsas, Andreas Loukas, Nikolaos Karalias, Michael M. Bronstein

37* Fitting summary statistics of neural data with a differentiable spiking network simulator    Poster    Guillaume Bellec, Shuqi Wang, Alireza Modirshanechi, Johanni Brea, Wulfram Gerstner

38* Learning where to learn: Gradient sparsity in meta and continual learning    Poster    Johannes Von Oswald, Dominic Zhao, Seijin Kobayashi, Simon Schug, Massimo Caccia, Nicolas Zucchet, Joao Sacramento

39* Information Directed Reward Learning for Reinforcement Learning    Poster    David Lindner, Matteo Turchetta, Sebastian Tschiatschek, Kamil Ciosek, and Andreas Krause

40* Prototypical Cross-Attention Networks for Multiple Object Tracking and Segmentation    Spotlight    Lei Ke, Xia Li, Martin Danelljan, Yu-Wing Tai, Chi-Keung Tang, Fisher Yu

41* STORM+: Fully Adaptive SGD with Recursive Momentum for Nonconvex Optimization    Poster    Kfir Y. Levy, Ali Kavis, Volkan Cevher

42* Repulsive Deep Ensembles are Bayesian    Spotlight    Francesco D’Angelo, Vincent Fortuin

43* Credit assignment in neural networks through Deep Feedback Control    Spotlight    Alexander Meulemans, Matilde Tristany Farinha, Javier García Ordóñez, Pau Vilimelis Aceituno, João Sacramento, Benjamin F. Grewe

44* The staircase property: how hierarchical structure can guide deep learning    Poster    E. Abbe, E. Boix-Adsera, M. Brenner, G. Bresler, D. Nagarj

45* On the Power of Differentiable Learning versus PAC and SQ Learning    Spotlight    E. Abbe, P. Kamath, E. Malach, C. Sandon, N. Srebro

46* Predicting future myocardial infarction from angiographies with deep learning    Other NeurIPS tracks (e.g. competition, benchmark…)    Dorina Thanou1 Ortal Senouf1 Omar Raita1 Emmanuel Abbé1 Pascal Frossard1 Farhang Aminfar2 Denise Auberson2 Nicolas Dayer2 David Meier2 Mattia Pagnoni2 Olivier Muller2 Stéphane Fournier2 Thabodhan Mahendiran2

47* Subquadratic Overparameterization for Shallow Neural Networks    Poster    Chaehwan Song, Ali Ramezani-Kebrya, Thomas Pethick, Armin Eftekhari, Volkan Cevher

48* Nearly-Tight and Oblivious Algorithms for Explainable Clustering    Poster    Buddhima Gamlath · Xinrui Jia · Adam Polak · Ola Svensson

49* Risk-averse Heteroscedastic Bayesian Optimization     Poster    Anastasia_Makarova , Ilnura Usmanova, Ilija Bogunovic, Andreas Krause

*50 Convergence of adaptive algorithms for constrained weakly convex
optimization, Ahmet Alacaoglu, Yura Malitsky, Volkan Cevher


*51 Sifting through the noise: Universal first-order methods for
stochastic variational inequalities, K. Antonakopoulos, T. Pethick, A.
Kavis, P. Mertikopoulos and V. Cevher


*52 Tree in Tree: from Decision Trees to Decision Graphs, Bingzhao
Zhu, Mahsa Shoaran


*53 DNN-based Topology Optimisation: Spatial Invariance and Neural
Tangent Kernel, Benjamin Dupuis, Arthur Jacot


*54 Precise characterization of the prior predictive distribution of
deep ReLU networks, Lorenzo Noci, Gregor Bachmann, Kevin Roth,
Sebastian Nowozin, Thomas Hofmann

*55  Interpolation can hurt robust generalization even when there is no noise, Konstantin Donhauser, Alexandru Tifrea, Michael Aerni, Reinhard Heckel, Fanny Yang


* 56 The Effect of the Intrinsic Dimension on the Generalization of Quadratic Classifiers,Fabian Latorre, Leello Tadesse Dadi, Paul Rolland, Volkan Cevher


* 57 Recursive Bayesian Networks: Generalising and Unifying Probabilistic Context-Free Grammars and Dynamic Bayesian Networks, Robert Lieck & Martin Rohrmeier

Replay

Part I

Part 2