BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Finding what is invisible through computation -  Dr Rafal Mantiuk 
 - University of Cambridge\, Computer Laboratory
DTSTART:20160120T140000Z
DTEND:20160120T150000Z
UID:TALK63463@talks.cam.ac.uk
CONTACT:David Greaves
DESCRIPTION:The success of any kind of visual interface\, from a smartphon
 e screen\,\nto a head-mounted display\, relies on the fact that the eye is
  rather\nbad in seeing the artefacts of those technologies. For example\, 
 we do\nnot see the matrix of red-green-blue subpixels or that video is in\
 nfact a sequence of frames presented in succession. Despite that we\nrely 
 on the limitations of the visual system in any aspect of visual\ntechnolog
 ies\, our level of understanding and computational modelling\nof those lim
 itations is surprisingly limited.\n\nIn this talk I will propose how compu
 tation can help to integrate\npartial information we have on the limitatio
 ns of the visual system\nand build a comprehensive model of those limitati
 ons. The challenge\nhere is different than in most machine learning proble
 ms: the amount\nof data is very limited\, expensive to collect\, and impos
 sible to\npredict with simple models. Some of the interesting problems are
  how\nto reuse existing psychophysical data\, or how to automatically find
 \nand generate new measurement points\, which would be the most\nbeneficia
 l for model discovery. The complexity of the models leads to\ncomputationa
 lly demanding training methods\, which require HPC\nresources and efficien
 t parallel implementations.\n
LOCATION:Lecture Theatre 1\, Computer Laboratory
END:VEVENT
END:VCALENDAR