works on

From the 1 of 13 linked papers with an AI index.

activity
20242026
collaborators
Showing cs.LGShow all

7 papers · 1 filter

cs.LG2026

Equilibrium Training of Energy-Based Models with Parallel Trajectory Tempering

Nicolas Béreux, Aurélien Decelle, Cyril Furtlehner +1

The paper proposes Parallel Trajectory Tempering (PTT), a training method that keeps equilibrium sampling throughout learning of energy‑based models, enabling fast and stable train…

cs.LG2026

Distributional simplicity bias and effective convexity in Energy Based Models

Aurélien Decelle, Alfonso de Jesús Navas Gómez, Beatriz Seoane

Energy-based learning is a powerful framework for generative modelling, but its training is inherently non-convex, leading potentially to sensitivity to initialisation, poor local…

cs.LG2025

Fast training and sampling of Restricted Boltzmann Machines

Nicolas Béreux, Aurélien Decelle, Cyril Furtlehner +2

Restricted Boltzmann Machines (RBMs) are powerful tools for modeling complex systems and extracting insights from data, but their training is hindered by the slow mixing of Markov…

cs.LG2025

PRIVET: Privacy Metric Based on Extreme Value Theory

Antoine Szatkownik, Aurélien Decelle, Beatriz Seoane +6

Deep generative models are often trained on sensitive data, such as genetic sequences, health data, or more broadly, any copyrighted, licensed or protected content. This raises cri…

cs.LG2025

A theoretical framework for overfitting in energy-based modeling

Giovanni Catania, Aurélien Decelle, Cyril Furtlehner +1

We investigate the impact of limited data on training pairwise energy-based models for inverse problems aimed at identifying interaction networks. Utilizing the Gaussian model as t…

cs.LG2025

Cascade of phase transitions in the training of Energy-based models

Dimitrios Bachtis, Giulio Biroli, Aurélien Decelle +1

In this paper, we investigate the feature encoding process in a prototypical energy-based generative model, the Restricted Boltzmann Machine (RBM). We start with an analytical inve…