Accelerating convergence of diffusion models | Vidéo | Carmin.tv

00:00:00 / 00:00:00

Accelerating convergence of diffusion models

Apparaît dans la collection : New challenges in high-dimensional statistics / Statistique mathématique 2025

Score-based diffusion models, while achieving remarkable empirical performance, often suffer from low sampling speed, due to extensive function evaluations needed during the sampling phase. Despite a flurry of recent activities towards speeding up diffusion generative modeling in practice, theoretical underpinnings for acceleration techniques remain severely limited. In this talk, we discuss novel training-free algorithms to accelerate popular deterministic (i.e., DDIM) and stochastic (i.e., DDPM) samplers.

Informations sur la vidéo

Date de captation 15/12/2025
Date de publication 12/01/2026
Institut CIRM
Licence CC BY NC ND
Langue Anglais
Audience Chercheurs, Doctorants
Réalisateur(s) Luca Recanzone
Format MP4

Données de citation

DOI 10.24350/CIRM.V.20425803
Citer cette vidéo Wei, Yuting (15/12/2025). Accelerating convergence of diffusion models. CIRM. Audiovisual resource. DOI: 10.24350/CIRM.V.20425803
URL https://dx.doi.org/10.24350/CIRM.V.20425803

Domaine(s)

Bibliographie

LI, Gen, ZHOU, Yuchen, WEI, Yuting, et al. Faster Diffusion Models via Higher-Order Approximation. arXiv preprint arXiv:2506.24042, 2025. - https://doi.org/10.48550/arXiv.2506.24042
LI, Gen, HUANG, Yu, EFIMOV, Timofey, et al. Accelerating convergence of score-based diffusion models, provably. arXiv preprint arXiv:2403.03852, 2024. - https://doi.org/10.48550/arXiv.2403.03852

Codes MSC

62D05 Sampling theory, sample surveys

Dernières questions liées sur MathOverflow

Pour poser une question, votre compte Carmin.tv doit être connecté à mathoverflow

Poser une question sur MathOverflow

Copyright Carmin.tv 2026

Donner son avis