BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Equivariant Imaging: Unsupervised Learning in Inverse Problems - D
 ongdong Chen\, University of Edinburgh
DTSTART:20211208T130000Z
DTEND:20211208T140000Z
UID:TALK166765@talks.cam.ac.uk
CONTACT:J.W.Stevens
DESCRIPTION:Deep networks provide state-of-the-art performance in multiple
  imaging inverse problems ranging from medical imaging to computational ph
 otography. In various imaging problems\, we usually only have access to co
 mpressed measurements of the underlying signals\, hindering most learning-
 based strategies which usually require pairs of signals and associated mea
 surements for training. Learning only from compressed measurements is impo
 ssible in general\, as the compressed observations do not contain informat
 ion outside the range of the forward sensing operator. In this talk\, I wi
 ll introduce a new end-to-end self-supervised framework\, called Equivaria
 nt Imaging (EI) that overcomes this limitation by exploiting the equivaria
 nces present in natural signals. Our proposed learning strategy performs a
 s well as fully supervised methods. Experiments demonstrate the potential 
 of this framework on inverse problems including sparse-view X-ray computed
  tomography\, accelerated MRI\, and image inpainting.\n\n*Join Zoom Meetin
 g*\nhttps://maths-cam-ac-uk.zoom.us/j/96932950870?pwd=b1o2c2UxckVITlRlazJz
 Y0laRmVHZz09\nMeeting ID: 969 3295 0870\nPasscode: DRHjehPj
LOCATION:Virtual (see abstract for Zoom link)
END:VEVENT
END:VCALENDAR