BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20181221T160730Z
LOCATION:D171
DTSTART;TZID=America/Chicago:20181114T103000
DTEND;TZID=America/Chicago:20181114T110000
UID:submissions.supercomputing.org_SC18_sess271_exforum105@linklings.com
SUMMARY:Productive and Performant AI Platforms of the Future
DESCRIPTION:Exhibitor Forum\nMachine Learning\n\nProductive and Performant
  AI Platforms of the Future\n\nSukumar\n\nThus far, contributions to hardw
 are and software tools for advanced analytics and artificial intelligence 
 has come from the commodity/cloud computing community. In this talk, we sh
 are exciting results from efforts that ported software frameworks such as 
 Apache Spark and TensorFlow onto high performance computing (HPC) hardware
  to make the case that HPC-approaches future-proof AI investments. We will
  demonstrate performance gains from combining a HPC interconnect with algo
 rithmic cleverness using communication collectives for graph analytics, de
 ep learning and matrix methods – all components of the modern data science
  and enterprise AI workflows. Based on experience from several use-cases, 
 we will argue how HPC futureproofs investments for AI journey – particular
 ly around emerging requirements around (I) non-traditional data (graphs, m
 edical imagery, genomic sequences); (ii) building custom domain-specific m
 odels with hyper-parameter learning; (iii) the need for ensemble and model
  parallelism (iv) latency on edge-devices and training cadence with custom
  processors.
URL:https://sc18.supercomputing.org/presentation/?id=exforum105&sess=sess2
 71
END:VEVENT
END:VCALENDAR