Learning Energy-Based Models from Stochastic Interpolants using Spatiotemporal Differences

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Yu, Hanlin, OuYang, RuiKang, Kaushik, Partha, Klami, Arto, Gutmann, Michael U., Chehab, Omar
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914604501696512
author Yu, Hanlin
OuYang, RuiKang
Kaushik, Partha
Klami, Arto
Gutmann, Michael U.
Chehab, Omar
author_facet Yu, Hanlin
OuYang, RuiKang
Kaushik, Partha
Klami, Arto
Gutmann, Michael U.
Chehab, Omar
contents Learning an energy-based model from data samples is a central problem in machine learning. Many recent and popular methods, such as denoising score matching for training energy-based diffusion models, use stochastic interpolants to corrupt data samples at different noise levels indexed by a time variable. This defines a joint density over both the data space and time, and most methods learn its energy through either spatial or temporal differences. We identify distinct failure modes for both of these approaches. To solve them, we propose Spatiotemporal Noise-Contrastive Estimation (stNCE), a framework for learning the energy through joint spatiotemporal differences. stNCE unifies many existing methods and leads to new training objectives. Experiments on images and molecules demonstrate performance competitive with state-of-the-art density estimation methods.
format Preprint
id arxiv_https___arxiv_org_abs_2605_26850
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Learning Energy-Based Models from Stochastic Interpolants using Spatiotemporal Differences
Yu, Hanlin
OuYang, RuiKang
Kaushik, Partha
Klami, Arto
Gutmann, Michael U.
Chehab, Omar
Machine Learning
Learning an energy-based model from data samples is a central problem in machine learning. Many recent and popular methods, such as denoising score matching for training energy-based diffusion models, use stochastic interpolants to corrupt data samples at different noise levels indexed by a time variable. This defines a joint density over both the data space and time, and most methods learn its energy through either spatial or temporal differences. We identify distinct failure modes for both of these approaches. To solve them, we propose Spatiotemporal Noise-Contrastive Estimation (stNCE), a framework for learning the energy through joint spatiotemporal differences. stNCE unifies many existing methods and leads to new training objectives. Experiments on images and molecules demonstrate performance competitive with state-of-the-art density estimation methods.
title Learning Energy-Based Models from Stochastic Interpolants using Spatiotemporal Differences
topic Machine Learning
url https://arxiv.org/abs/2605.26850