BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230831T095747Z
LOCATION:Seehorn
DTSTART;TZID=Europe/Stockholm:20230628T123000
DTEND;TZID=Europe/Stockholm:20230628T130000
UID:submissions.pasc-conference.org_PASC23_sess121_msa239@linklings.com
SUMMARY:AI for Large-Scale High-Dynamic Range Image Reconstruction in Radi
 o Astronomy
DESCRIPTION:Minisymposium\n\nYves Wiaux and Arwa Dabbech (Heriot-Watt Univ
 ersity)\n\nEndowing modern radio-interferometric (RI) telescopes with an a
 cute vision requires solving large-scale high-dynamic range inverse proble
 ms for image formation. This calls for sophisticated computational imaging
  algorithms injecting an image model into the observed data. We discuss tw
 o AI-based solutions to break the barrier towards joint precision and scal
 ability of RI imaging algorithms. The plug-and-play algorithm AIRI (arXiv:
 2210.16060) entails learning a prior image model by training a deep neural
  network (DNN) as a denoiser, and substituting it for the handcrafted, oft
 en sub-iterative, regularisation operator of an optimisation algorithm. AI
 RI is encapsulated in a parallel framework, equipped with automated functi
 onalities for faceting its denoiser and decomposing the RI measurement ope
 rator into blocks. AIRI was demonstrated on gigabyte-sized data to deliver
  similar precision to its optimisation counterpart SARA, at reduced comput
 ational cost. Yet AIRI's highly iterative nature hinders scalability to ex
 treme dimension. R2D2 (arXiv:2210.16060) is a learned version of the well-
 known matching pursuit algorithm. It delivers a reconstruction within few 
 iterations, built as a series of residual images output by DNNs from data 
 residuals. Preliminary results leveraging the same parallel measurement op
 erator, suggest that R2D2 achieves similar precision to AIRI at a fraction
  of the cost, opening the door to real-time RI imaging.\n\nDomain: Compute
 r Science, Machine Learning, and Applied Mathematics 
\n\nSession Chair: M
 atthieu Simeoni (EPFL)
END:VEVENT
END:VCALENDAR