CIRM - Videos & books Library - Biased Monte Carlo sampling in RBMs

Post-edited

Auteurs : Seoane, Beatriz (Auteur de la conférence)
CIRM (Editeur )

Loading the player...

Generative models Restricetd Boltzmann machine Training an RBM Interpretability of machine learning Sampling problems RBM operational regimes Mixing time Metastability problems Biased sampling

Résumé : RBMs are generative models capable of fitting complex dataset's probability distributions. Thanks to their simple structure, they are particularly well suited for interpretability and pattern extraction, a feature particularly appealing for scientific use. In this talk, we show that RBMs operate in two distinct regimes, depending on the procedure followed to estimate the log-likelihood gradient during the training. Short sampling times fit machines that are trained to reproduce exactly the dynamics followed to train them, long samplings (as compared to the MCMC mixing time) are need to learn a good model for the data. The non-equilibrium regime should be used to generate high quality samples in short learning and sampling times, but cannot be used to extract the unnormalized data probability of the data necessary for interpretability. In practice, it is hard to extract good equilibrium models for structured datasets (which is the typical case in biological applications) due to a divergence of the Monte Carlo mixing times. In this work, we show this barrier can be surmounted using biased Monte Carlo methods.

Mots-Clés : restricted Boltzmann machines; energy based models; generative models; unsupervised learning; mixing time

Codes MSC :
65C05 - Monte Carlo methods
82C44 - Dynamics of disordered systems (random Ising systems, etc.)
68T07 - Artificial neural networks and deep learning

Ressources complémentaires :
https://www.cirm-math.fr/RepOrga/2389/Slides/Seoane.pdf

Informations sur la Vidéo

Réalisateur :

Hennenfent, Guillaume

Langue :

Date de Publication :

Date de Captation :

Sous Collection :

Catégorie arXiv :

Domaine(s) :

Format :

Durée :

Audience :

Download :

https://videos.cirm-math.fr/2021-09-30_Seoane.mp4

Informations sur la Rencontre

Nom de la Rencontre : On Future Synergies for Stochastic and Learning Algorithms / Sur les synergies futures autour des algorithmes d'apprentissage et stochastiques
Organisateurs de la Rencontre : Durmus, Alain ; Michel, Manon ; Roberts, Gareth ; Zdeborova, Lenka
Dates : 27/09/2021 - 01/10/2021
Année de la rencontre : 2021
URL de la Rencontre : https://conferences.cirm-math.fr/2389.html

Données de citation

DOI : 10.24350/CIRM.V.19817903
Citer cette vidéo: Seoane, Beatriz (2021). Biased Monte Carlo sampling in RBMs. CIRM. Audiovisual resource. doi:10.24350/CIRM.V.19817903
URI : http://dx.doi.org/10.24350/CIRM.V.19817903

Voir Aussi

[Multi angle] MCMC for systems with glassy dynamics / Auteur de la conférence Berthier, Ludovic.
[Multi angle] Irreversible Monte Carlo methods for particle simulations / Auteur de la conférence Maggs, Anthony.
[Multi angle] Molecular dynamics algorithms: numerical and mathematical analysis / Auteur de la conférence Lelièvre, Tony.
[Multi angle] Coupling rare event algorithms and deep neural network to predict extreme heat waves / Auteur de la conférence Bouchet, Freddy.

Bibliographie

DECELLE, Aurélien, FURTLEHNER, Cyril, et SEOANE, Beatriz. Equilibrium and non-Equilibrium regimes in the learning of Restricted Boltzmann Machines. arXiv preprint arXiv:2105.13889, 2021. - https://arxiv.org/abs/2105.13889

DECELLE, Aurélien, FISSORE, Giancarlo, et FURTLEHNER, Cyril. Thermodynamics of restricted Boltzmann machines and related learning dynamics. Journal of Statistical Physics, 2018, vol. 172, no 6, p. 1576-1608. - http://dx.doi.org/10.1007/s10955-018-2105-y

DECELLE, Aurélien et FURTLEHNER, Cyril. Exact training of Restricted Boltzmann machines on intrinsically low dimensional data. arXiv preprint arXiv:2103.10755, 2021. - https://arxiv.org/abs/2103.10755

MARTIN-MAYOR, V., SEOANE, B., et YLLANES, D. Tethered Monte Carlo: Managing rugged free-energy landscapes with a Helmholtz-potential formalism. Journal of Statistical Physics, 2011, vol. 144, no 3, p. 554-596. - http://dx.doi.org/10.1007/s10955-011-0261-4

FERNÁNDEZ, L. A., MARTÍN-MAYOR, Víctor, SEOANE, B., et al. Equilibrium fluid-solid coexistence of hard spheres. Physical review letters, 2012, vol. 108, no 16, p. 165701. - http://dx.doi.org/10.1103/PhysRevLett.108.165701

Seoane.jpg

Sélection Signaler une erreur

TUTELLES

PARTENAIRES

Destination de la recherche

Raccourcis

Voir Aussi