BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260114T163656Z
LOCATION:Meeting Room C4.9+C4.10\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231214T121500
DTEND;TZID=Australia/Melbourne:20231214T122500
UID:siggraphasia_SIGGRAPH Asia 2023_sess170_papers_878@linklings.com
SUMMARY:Example-Based Sampling with Diffusion Models
DESCRIPTION:Bastien Doignies (Université Claude Bernard Lyon, CNRS); Nicol
 as Bonneel, David Coeurjolly, and Julie Digne (CNRS, LIRIS); Loïs Paulin (
 Université Claude Bernard Lyon / CNRS, Adobe); and Jean-Claude Iehl and Vi
 ctor Ostromoukhov (Université Claude Bernard Lyon, CNRS)\n\nMuch effort ha
 s been put into developing samplers with specific properties, such as prod
 ucing blue noise, low-discrepancy, lattice or Poisson disk samples. These 
 samplers can be slow if they rely on optimization processes, may rely on a
  wide range of numerical methods, are not always differentiable. The succe
 ss of recent diffusion models for image generation suggests that these mod
 els could be appropriate for learning how to generate point sets from exam
 ples. However, their convolutional nature makes these methods impractical 
 for dealing with scattered data such as point sets. We propose a generic w
 ay to produce 2-d point sets imitating existing samplers from observed poi
 nt sets using a diffusion model. We address the problem of convolutional l
 ayers by leveraging neighborhood information from an optimal transport mat
 ching to a uniform grid, that allows us to benefit from fast convolutions 
 on grids, and to support the example-based learning of non-uniform samplin
 g patterns. We demonstrate how the differentiability of our approach can b
 e used to optimize point sets to enforce properties.\n\nRegistration Categ
 ory: Full Access\n\nSession Chair: Xiangyu Xu (Xi'an Jiaotong University)\
 n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_878&sess=sess170
END:VEVENT
END:VCALENDAR