BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Canada/Pacific
BEGIN:DAYLIGHT
DTSTART:20190310T030000
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:PDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20191103T010000
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:PST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190712T045203Z
UID:E0BB6015-F6B8-4293-B2FB-96F3AD0489BE
DTSTART;TZID=Canada/Pacific:20190711T123000
DTEND;TZID=Canada/Pacific:20190711T133000
DESCRIPTION:The KeOps library lets you compute generic reductions of large 
 2d arrays whose entries are given by a mathematical formula. It is perfect
 ly suited to the computation of convolutions (or more generally to Kernel 
 dot products) and the associated gradients (with an automatic differentiat
 ion engine).\n\nKeOps is fast as it allows you to compute Gaussian convolu
 tion up to 40 times faster than a standard tensor algebra library that use
  GPU. KeOps is scalable and can be used on large data (typically from n=10
 ^3 to n=10^7 number of rows/columns): it combines a tiled reduction scheme
  and works even when the full kernel matrix does not fit into the GPU memo
 ry. Finally\, KeOps is easy to use as it comes with its Matlab\, Python (N
 umPy or PyTorch) and R (coming soon) bindings.\n\nWeb site: http://www.ker
 nel-operations.io\n\nSpeaker(s): Dr. Benjamin Charlier\, \n\nRoom: ASB 109
 00\, Bldg: Applied Sciences Building\, 8888 University Drive\, School of E
 ngineering Science\, Burnaby\, British Columbia\, Canada\, V5A 1S6
LOCATION:Room: ASB 10900\, Bldg: Applied Sciences Building\, 8888 Universit
 y Drive\, School of Engineering Science\, Burnaby\, British Columbia\, Can
 ada\, V5A 1S6
ORGANIZER:ivan_bajic@ieee.org
SEQUENCE:3
SUMMARY:KeOps: Kernel Operations on the GPU\, with autodiff\, without memor
 y overflows
URL;VALUE=URI:https://events.vtools.ieee.org/m/201275
X-ALT-DESC:Description: <br /><p>The KeOps library lets you compute generic
  reductions of large 2d arrays whose entries are given by a mathematical f
 ormula. It is perfectly suited to the computation of convolutions (or more
  generally to Kernel dot products) and the associated gradients (with an a
 utomatic differentiation engine).</p>\n<p>KeOps is fast as it allows you t
 o compute Gaussian convolution up to 40 times faster than a standard tenso
 r algebra library that use GPU. KeOps is scalable and can be used on large
  data (typically from n=10^3 to n=10^7 number of rows/columns): it combine
 s a tiled reduction scheme and works even when the full kernel matrix does
  not fit into the GPU memory. Finally\, KeOps is easy to use as it comes w
 ith its Matlab\, Python (NumPy or PyTorch) and R (coming soon) bindings.</
 p>\n<p>Web site: http://www.kernel-operations.io</p>\n<p>&nbsp\;</p>
END:VEVENT
END:VCALENDAR