BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Machine learning tools for large tomographic inver
 se problems with limited training data - Thomas Bl
 umensath (University of Southampton)
DTSTART;TZID=Europe/London:20230331T095000
DTEND;TZID=Europe/London:20230331T104000
UID:TALK198283AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/198283
DESCRIPTION:X-ray tomographic inverse problems with limited me
 asurements require strong non-linear constraints. 
 Increasingly\, machine learning tools are used to 
 learn these constraints from large collections of 
 representative training data. When using X-ray tom
 ography to image manufactured components\, it is o
 ften beneficial to target the training data to the
  specific application\, as this can lead to very s
 trong constraints that will allow image reconstruc
 tion even if significant amounts of measurements a
 re missing. There are however two fundamental prob
 lems with this approach for real applications. Fir
 stly\, it is often difficult to collect sufficient
  training data to train the most advanced machine 
 learning models. Secondly\, the inverse problem is
  extremely large\, with billions of measurements u
 sed to estimate 3D images with billions of voxels.
  This further restricts the models that can be tra
 ined and used on most computing hardware. We here 
 report on the use of block based 3D image models a
 nd show how they can be trained on a single 3D ima
 ge. This approach can be used for image de-noising
  as well as a building block in an unrolled optimi
 sation algorithm to solve the tomographic inverse 
 problem.&nbsp\;
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:
END:VEVENT
END:VCALENDAR