BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Dynamical Grammars for Galaxy Image Recognition - Wayne Hayes - UC
  Irvine\, USA
DTSTART:20100526T131500Z
DTEND:20100526T141500Z
UID:TALK23759@talks.cam.ac.uk
CONTACT:Mateja Jamnik
DESCRIPTION:Data from sky images are large and growing.  The Sloan\nDigita
 l Sky Survey (SDSS) contains an estimated 1 million galaxy\nimages.  The L
 arge Synoptic Survey Telescope (LSST) is being built and\nwill scan the en
 tire sky repeatedly\, providing images of millions of\ngalaxies and petaby
 tes of data every night. The SuperNova Acceleration\nProbe (SNAP) is a pro
 posed orbiting satellite that will repeatedly map\nthe entire sky from ori
 bit\, providing images of perhaps billions of\ngalaxies. Unfortunately\, g
 iven an image of a spiral galaxy\, there does\nnot exist an automated visi
 on algorithm to even tell us which direction\nthe spiral arms wind\, much 
 less count them or provide any other\nquantitative information about them.
  To wit\, the largest galaxy\nclassification project is the Galaxy Zoo\, i
 n which thousands of human\nvolunteers classify images by eye over the web
 . Although valuable\, such\nhuman classifications will provide only limite
 d objective quantitative\nmeasurements\, and soon be overwhelmed with more
  data than humans can\nhandle. However\, such information would prove an i
 nvaluable source for\nastronomers and cosmologists to test current theorie
 s of galaxy\nformation and cosmic evolution (which can now be simulated wi
 th high\naccuracy on large computers\, producing copious predictions that 
 cannot\nbe tested due to a lack of objective\, quantitative observational 
 data).\nIn this talk\, I will report on preliminary results from dynamical
 \ngrammars and other machine learning and vision techniques to "parse"\nim
 ages of galaxies\, starting us on the road towards producing\nquantitative
  data that will be useful for astronomers to test\nscientific theories. Th
 is work is in collaboration with Darren Davis\,\nEric Mjolsness\, Aaron Ba
 rth\, of UC Irvine.
LOCATION:Lecture Theatre 1\, Computer Laboratory
END:VEVENT
END:VCALENDAR