00:00:00 / 00:00:00

Gradient flows for sampling and their deterministic interacting particle approximations

De Dejan Slepčev

Apparaît dans la collection : Aggregation-Diffusion Equations & Collective Behavior: Analysis, Numerics and Applications / Conférence Chaire Jean Morlet: Equations d'agrégation-diffusion et comportement collectif: Analyse, schémas numériques et applications

Motivated by the task of sampling measures in high dimensions we will discuss a number of gradient flows in the spaces of measures, including the Wasserstein gradient flows of Maximum Mean Discrepancy and Hellinger gradient flows of relative entropy, the Stein Variational Gradient Descent and a new projected dynamic gradient flows. For all the flows we will consider their deterministic interacting-particle approximations. The talk is highlight some of the properties of the flows and indicate their differences. In particular we will discuss how well can the interacting particles approximate the target measures.The talk is based on joint works wit Anna Korba, Lantian Xu, Sangmin Park, Yulong Lu, and Lihan Wang.

Informations sur la vidéo

Données de citation

  • DOI 10.24350/CIRM.V.20160203
  • Citer cette vidéo Slepčev, Dejan (11/04/2024). Gradient flows for sampling and their deterministic interacting particle approximations. CIRM. Audiovisual resource. DOI: 10.24350/CIRM.V.20160203
  • URL https://dx.doi.org/10.24350/CIRM.V.20160203

Bibliographie

  • LU, Yulong, SLEPČEV, Dejan, et WANG, Lihan. Birth–death dynamics for sampling: global convergence, approximations and their asymptotics. Nonlinearity, 2023, vol. 36, no 11, p. 5731. - http://dx.doi.org/10.1088/1361-6544/acf988
  • PARK, Sangmin et SLEPČEV, Dejan. Geometry and analytic properties of the sliced Wasserstein space. arXiv preprint arXiv:2311.05134, 2023. - https://arxiv.org/abs/2311.05134
  • XU, Lantian, KORBA, Anna, et SLEPCEV, Dejan. Accurate quantization of measures via interacting particle-based optimization. In : International Conference on Machine Learning. PMLR, 2022. p. 24576-24595. - https://proceedings.mlr.press/v162/xu22d.html

Dernières questions liées sur MathOverflow

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

Poser une question sur MathOverflow




Inscrivez-vous

  • Mettez des vidéos en favori
  • Ajoutez des vidéos à regarder plus tard &
    conservez votre historique de consultation
  • Commentez avec la communauté
    scientifique
  • Recevez des notifications de mise à jour
    de vos sujets favoris
Donner son avis
Loading…
Loading the web debug toolbar…
Attempt #