Publications associées à Network Loss Inference with Second Order Statistics of End-to-End Flows

Random matrix methods for high-dimensional machine learning models

In the rapidly evolving landscape of machine learning research, neural networks stand out with their ever-expanding number of parameters and reliance on increasingly large datasets. The financial cost and computational resources required for the training p ...

EPFL2024

Tiers of joy? Reserve tiering and bank behavior in a negative-rate environment

Andreas Fuster

Negative interest rate regimes typically involve reserve tiering to exempt a portion of bank reserves from negative rates. We study the effects on bank behavior of a large and unanticipated change in reserve tiering by the Swiss National Bank that generate ...

2024

Spatial Modeling for Building Design Evaluation: from Visual Landscape Quality Assessment to Devaluation Risk Estimation

Adam Robert Swietek

Zoning reform is a crucial tool for cities to adapt to contemporary challenges. However, its implementation remains challenging. Property owners, with a vested interest in the value of their neighborhoods, are sensitive to local developments and the potent ...

EPFL2024

Validation of semi-analytical, semi-empirical covariance matrices for two-point correlation function for early DESI data

Cheng Zhao

We present an extended validation of semi-analytical, semi-empirical covariance matrices for the two-point correlation function (2PCF) on simulated catalogs representative of luminous red galaxies (LRGs) data collected during the initial 2 months of operat ...

OXFORD UNIV PRESS2023

Euclid preparation: XXVIII. Forecasts for ten different higher-order weak lensing statistics

Frédéric Courbin, Fabio Finelli, Richard Massey, Maurizio Martinelli, Jean-Luc Starck, Gianluca Castignani, Marcello Farina, Austin Chandler Peel, Yi Wang

Recent cosmic shear studies have shown that higher-order statistics (HOS) developed by independent teams now outperform standard two-point estimators in terms of statistical precision thanks to their sensitivity to the non-Gaussian features of large-scale ...

EDP SCIENCES S A2023

On the symmetries in the dynamics of wide two-layer neural networks

Lénaïc Chizat

We consider the idealized setting of gradient flow on the population risk for infinitely wide two-layer ReLU neural networks (without bias), and study the effect of symmetries on the learned parameters and predictors. We first describe a general class of s ...

AMER INST MATHEMATICAL SCIENCES-AIMS2023

DESI mock challenge Halo and galaxy catalogues with the bias assignment method

Yu Yu, Cheng Zhao, Ginevra Favole

Context. We present a novel approach to the construction of mock galaxy catalogues for large-scale structure analysis based on the distribution of dark matter halos obtained with effective bias models at the field level. Aims. We aim to produce mock galaxy ...

EDP SCIENCES S A2023

Predicting the stressed expected loss of large US banks

Eric Jondeau, Amir Hossein Khalilzadeh Naghneh

We develop a methodology to measure the expected loss of commercial banks in a market downturn, which we call stressed expected loss (SEL). We simulate a market downturn as a negative shock on interest rate and credit market risk factors that reflect the b ...

ELSEVIER2022

Stochastic distributed learning with gradient quantization and double-variance reduction

Sebastian Urban Stich, Konstantin Mishchenko

We consider distributed optimization over several devices, each sending incremental model updates to a central server. This setting is considered, for instance, in federated learning. Various schemes have been designed to compress the model updates in orde ...

TAYLOR & FRANCIS LTD2022

The DESI N-body Simulation Project - II. Suppressing sample variance with fast simulations

Yu Yu, Andrei Variu, Cheng Zhao, Hanyu Zhang

Dark Energy Spectroscopic Instrument (DESI) will construct a large and precise three-dimensional map of our Universe. The survey effective volume reaches similar to 20 h(-3) Gpc(3). It is a great challenge to prepare high-resolution simulations with a much ...

OXFORD UNIV PRESS2022

Unbiased Monte Carlo cluster updates with autoregressive neural networks

Giuseppe Carleo, Dian Wu, Riccardo Rossi

Efficient sampling of complex high-dimensional probability distributions is a central task in computational science. Machine learning methods like autoregressive neural networks, used with Markov chain Monte Carlo sampling, provide good approximations to s ...

AMER PHYSICAL SOC2021

Euclid : Effects of sample covariance on the number counts of galaxy clusters

Georges Meylan, Richard Massey, Yi Wang

Aims. We investigate the contribution of shot-noise and sample variance to uncertainties in the cosmological parameter constraints inferred from cluster number counts, in the context of the Euclid survey. ...

EDP SCIENCES S A2021

Fast ion transport phenomena in turbulent toroidal plasmas

Fabian Manke

In the field of plasma physics, suprathermal ions are encountered e.g. in astrophysical jets, the solar wind, as well as fusion plasmas, where they originate from neutral beam injection or fusion reactions. One aspect of shared interest with astrophysical ...

EPFL2020

Frequency domain theory for functional time series: Variance decomposition and an invariance principle

Neda Mohammadi Jouzdani

This paper is concerned with frequency domain theory for functional time series, which are temporally dependent sequences of functions in a Hilbert space. We consider a variance decomposition, which is more suitable for such a data structure than the varia ...

INT STATISTICAL INST2020

Generalization error in high-dimensional perceptrons: Approaching Bayes error with convex optimization

Florent Gérard Krzakala, Lenka Zdeborová

We consider a commonly studied supervised classification of a synthetic dataset whose labels are generated by feeding a one-layer non-linear neural network with random iid inputs. We study the generalization performances of standard classifiers in the high ...

Curran Associates, Inc.2020

Cover: A Cluster-Based Variance Reduced Method For Online Learning

Ali H. Sayed, Bicheng Ying, Kun Yuan

In this paper, we develop a stochastic-gradient learning algorithm for situations involving streaming data that arise from an underlying clustered structure. In such settings, the variance of gradient noise can be decomposed into the in-cluster variance si ...

IEEE2019

Applying dimension reduction to EEG data by Principal Component Analysis reduces the quality of its subsequent Independent Component decomposition

Fiorenzo Artoni

Independent Component Analysis (ICA) has proven to be an effective data driven method for analyzing EEG data, separating signals from temporally and functionally independent brain and non-brain source processes and thereby increasing their definition. Dime ...

2018

IT Investment and its impacts : does IT matter during crisis, during financing and in Innovation?

Shameem Hasan

This thesis empirically explored the impacts of IT investment on three different scenarios under the common denominator / threads of IT investments. It comprises three different essays on the impacts of IT investments. IT investments and its impact on many ...

EPFL2018

The regularized tau estimator: A robust and efficient solution to ill-posed linear inverse problems

Martin Vetterli, Marta Martinez-Camara, Benjamin Bejar Haro

Linear inverse problems are ubiquitous. Often the measurements do not follow a Gaussian distribution. Additionally, a model matrix with a large condition number can complicate the problem further by making it ill-posed. In this case, the performance of pop ...

2017

Contributions to Modelling Extremes of Spatial Data

Linda Frossard

The increasing interest in using statistical extreme value theory to analyse environmental data is mainly driven by the large impact extreme events can have. A difficulty with spatial data is that most existing inference methods for asymptotically justifie ...

EPFL2017