BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//NextToU//EN
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:7a3813b4-3415-44e1-ad0c-e34eeb3aaab7@nexttoyou
DTSTAMP:20261105T152501Z
DTSTART:20261105T170000Z
DTEND:20261105T180000Z
SUMMARY:Distinguished Seminar in Computational Science and Engineering
LOCATION:Building 45 (MIT Stephen A. Schwarzman College of Computing) 45-4
 32\, 51 Vassar Street\, Cambridge\, MA 02139-4307
DESCRIPTION:Distinguished Seminar in Computational Science and Engineering
 \n\nNovember 5\, 2026\, 12-1PM\n\n45-432 in Building 45 and Zoom Webinar\
 n\nMachine learning for computational imaging: What did we learn? What di
 d the machines learn?\n\nGeorge Barbastathis\n\nRalph E. and Eloise F. Cr
 oss Professor of Manufacturing\, Professor of Mechanical Engineering\, MI
 T\n\nAbstract:\n\nFor the past decade\, my group has been working on mach
 ine learning for computational imaging and\, more specifically\, supervis
 ed training of regularizers for ill-posed and ill-conditioned inverse pro
 blems. The notion of data-driven regularization has been around since at 
 least the invention of learned sparse codes\, or “dictionaries\,” by Olsh
 ausen and Field in 1996. Learning the code through a deep neural network\
 , as pointed out in 2010 by Gregor and Lecun\, narrows the regularizer on
 -demand. Thus\, improvements should be expected in terms of resilience to
  noise and incomplete measurements. Indeed\, circa 2017~18\, Unser’s grou
 p at EPFL devised a cascaded neural proximal gradient scheme for linear t
 omography with reduced exposure dose\; Ozcan’s group at UCLA showed that 
 a low-NA microscope objective can capture features of comparable fidelity
  as high-NA\; and my group at MIT discovered that a Gerchberg-Saxton-Fien
 up module followed by a deep neural network can retrieve the phase of the
  complex field robustly from highly noisy intensity measurements.\nUsing 
 neural networks for imaging in various contexts is now widespread. What h
 ave we le\n\nhttp://next-to-u.app/events/7a3813b4-3415-44e1-ad0c-e34eeb3a
 aab7
URL:http://next-to-u.app/events/7a3813b4-3415-44e1-ad0c-e34eeb3aaab7
END:VEVENT
END:VCALENDAR