BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An Introduction to Stochastic Interpolants - Professor Jose Miguel
  Hernandez Lobato and Fran Castro
DTSTART:20251022T100000Z
DTEND:20251022T113000Z
UID:TALK239803@talks.cam.ac.uk
CONTACT:Xianda Sun
DESCRIPTION:In this first reading group of the year\, we will be giving an
  introduction to the framework of stochastic interpolants and showing thei
 r connection to flow matching and diffusion. Stochastic interpolants enabl
 es the use of a broad class of continuous-time stochastic processes to bri
 dge any two probability density functions exactly in finite time. These in
 terpolants are built by combining data from the two prescribed densities w
 ith an additional latent variable. The time-dependent density function of 
 the interpolant satisfies a transport equation as well as a family of forw
 ard and backward Fokker-Planck equations with a tunable diffusion coeffici
 ent. This leads to both deterministic and stochastic generative models bas
 ed on probability flow equations or stochastic differential equations with
  an adjustable level of noise.\n\nIn the first part of the tutorial we wil
 l have a look at the theory behind stochastic interpolants. In the last pa
 rt of the tutorial\, we will describe an application of the stochastic int
 erpolants framework to design a new sampling based method that seems compe
 titive with state of the art techniques like parallel tempering.\n\nReadin
 g material (not necessary to read before the session):\n\n* Building Norma
 lizing Flows with Stochastic Interpolants\nMichael S. Albergo\, Eric Vande
 n-Eijnden\nhttps://arxiv.org/abs/2209.15571\n\n* Stochastic Interpolants: 
 A Unifying Framework for Flows and Diffusions\nMichael S. Albergo\, Nichol
 as M. Boffi\, Eric Vanden-Eijnden\nhttps://arxiv.org/abs/2303.08797\n\n* F
 low Matching for Generative Modeling\nYaron Lipman\, Ricky T. Q. Chen\, He
 li Ben-Hamu\, Maximilian Nickel\, Matt Le\nhttps://arxiv.org/abs/2210.0274
 7
LOCATION:Cambridge University Engineering Department\, CBL Seminar room BE
 4-38.
END:VEVENT
END:VCALENDAR
