Fast and Functional structured data generator - Analytical Genomics Access content directly
Conference Papers Year : 2023

Fast and Functional structured data generator

Abstract

In this study, we address the challenge of using energy-based models to produce high-quality, label-specific data in complex structured datasets. Traditional training methods encounter difficulties due to inefficient Markov chain Monte Carlo mixing, which affects the diversity of synthetic data and increases generation times. To address these issues, we use a novel training algorithm that exploits non-equilibrium MCMC effects. This approach improves the model's ability to correctly classify samples and generate high-quality samples in only a few sampling steps. The effectiveness of this method is demonstrated learning three datasets with Restricted Boltzmann Machines: handwritten digits for visualization, a human mutation genome dataset classified by continental origin, and sequences of an enzyme protein family categorized by experimental biological function.
Fichier principal
Vignette du fichier
56_fast_and_functional_structured.pdf (7.26 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04342214 , version 1 (13-12-2023)

Licence

Attribution

Identifiers

  • HAL Id : hal-04342214 , version 1

Cite

Alessandra Carbone, Aurélien Decelle, Lorenzo Rosset, Beatriz Seoane. Fast and Functional structured data generator. ICML 2023 - Workshop on Structured Probabilistic Inference & Generative Modeling, Jul 2023, Honolulu, United States. ⟨hal-04342214⟩
78 View
15 Download

Share

Gmail Facebook X LinkedIn More