Publications associées à Perturbative construction of mean-field equations in extensive-rank matrix factorization and denoising

Publications associées (13)

Optimal Control under Uncertainty: From Regret Minimization to Distributional Robustness

Recent years have witnessed a growing integration of automatic control into a wide range of cyber-physical systems, including electrical grids and autonomous transportation networks. These systems operate in uncertain and dynamic environments, exhibit comp ...

EPFL2025

Advancing understanding and practical performance of machine learning interatomic potentials

Sergey Pozdnyakov

Atomistic machine learning has revolutionized atomistic modeling. Traditionally, molecular dynamics simulations relied on classical force fields or electronic structure calculations like Density Functional Theory. Classical force fields are fast but less p ...

EPFL2025

Optimization Algorithms for Decentralized, Distributed and Collaborative Machine Learning

Anastasiia Koloskova

Distributed learning is the key for enabling training of modern large-scale machine learning models, through parallelising the learning process. Collaborative learning is essential for learning from privacy-sensitive data that is distributed across various ...

EPFL2024

High-dimensional Asymptotics of Denoising Autoencoders

Lenka Zdeborová, Hugo Chao Cui

We address the problem of denoising data from a Gaussian mixture using a twolayer non-linear autoencoder with tied weights and a skip connection. We consider the high-dimensional limit where the number of training samples and the input dimension jointly te ...

2023

MILP-based discrete sizing and topology optimization of truss structures: new formulation and benchmarking

Corentin Jean Dominique Fivet, Gennaro Senatore, Jan Friedrich Georg Brütting

Discrete sizing and topology optimization of truss structures subject to stress and displacement constraints has been formulated as a Mixed-Integer Linear Programming (MILP) problem. The computation time to solve a MILP problem to global optimality via a b ...

2022

On Linear Optimization over Wasserstein Balls

Daniel Kuhn, Wolfram Wiesemann

Wasserstein balls, which contain all probability measures within a pre-specified Wasserstein distance to a reference measure, have recently enjoyed wide popularity in the distributionally robust optimization and machine learning communities to formulate an ...

2022

Binary Perceptron: Efficient Algorithms Can Find Solutions in a RareWell-Connected Cluster

Emmanuel Abbé

It was recently shown that almost all solutions in the symmetric binary perceptron are isolated, even at low constraint densities, suggesting that finding typical solutions is hard. In contrast, some algorithms have been shown empirically to succeed in fin ...

ASSOC COMPUTING MACHINERY2022

Algebraic Homotopy Interleaving Distance

Nicolas Michel Berkouk

The theory of persistence, which arises from topological data analysis, has been intensively studied in the one-parameter case both theoretically and in its applications. However, its extension to the multi-parameter case raises numerous difficulties, wher ...

SPRINGER INTERNATIONAL PUBLISHING AG2021

Implicit Bias of SGD for Diagonal Linear Networks: a Provable Benefit of Stochasticity

Nicolas Henri Bernard Flammarion, Scott William Pesme, Loucas Pillaud-Vivien

Understanding the implicit bias of training algorithms is of crucial importance in order to explain the success of overparametrised neural networks. In this paper, we study the dynamics of stochastic gradient descent over diagonal linear networks through i ...

2021

Generalized resolution matrix for neutron spin-echo three-axis spectrometers

Thomas Keller, Felix Groitl

This article describes the energy resolution of spin-echo three-axis spectrometers (SE-TASs) by a compact matrix formalism. SE-TASs allow one to measure the line widths of elementary excitations in crystals, such as phonons and magnons, with an energy reso ...

INT UNION CRYSTALLOGRAPHY2018

AC OPF in radial distribution networks – Part II: An augmented Lagrangian-based OPF algorithm, distributable via primal decomposition

Mario Paolone, Jean-Yves Le Boudec, Dan-Cristian Tomozei, Konstantina Christakou

In the first part of this two-part paper we show that the branch-flow convexification of the OPF problem is not exact and that the ADMM-based decomposition of the OPF fails to converge in specific scenarios. Therefore, there is a need to develop algorithms ...

Elsevier2017

A novel MILP approach for simultaneous optimization of water and energy: Application to a Canadian softwood Kraft pulping mill

François Maréchal, Maziar Kermani

An optimization methodology based on Mixed Integer Linear Programming (MILP) has been developed for simultaneous optimization of water and energy (SOWE) in industrial processes. The superstructure integrates non-water process thermal streams and optimizes ...

2017

On solving LPN using BKW and variants Implementation and Analysis

Serge Vaudenay, Sonia Mihaela Bogos, Florian Tramèr

The Learning Parity with Noise problem (LPN) is appealing in cryptography as it is considered to remain hard in the post-quantum world. It is also a good candidate for lightweight devices due to its simplicity. In this paper we provide a comprehensive anal ...

Springer2016