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:20211207T055405Z
LOCATION:Online
DTSTART;TZID=America/Chicago:20211119T105000
DTEND;TZID=America/Chicago:20211119T111000
UID:submissions.supercomputing.org_SC21_sess334_ws_lasalss104@linklings.co
 m
SUMMARY:Passel: Improved Scalability and Efficiency of Distributed SVM Usi
 ng a Cacheless PGAS Migrating Thread Architecture
DESCRIPTION:Workshop\n\nPassel: Improved Scalability and Efficiency of Dis
 tributed SVM Using a Cacheless PGAS Migrating Thread Architecture\n\nPage,
  Kogge\n\nStochastic Gradient Descent (SGD) is a valuable algorithm for la
 rge-scale machine learning, but has proven difficult to parallelize on con
 ventional architectures because of communication and memory access issues.
  The HogWild series of mixed logically distributed and physically multi-th
 readed algorithms overcomes these issues for problems with sparse characte
 ristics by using multiple local model vectors with asynchronous atomic upd
 ates. While this approach has proven effective for several reported exampl
 es, there are others, especially very sparse cases, that do not scale as w
 ell. This paper discusses an SGD Support Vector Machine (SVM) on a cachele
 ss migrating thread architecture using the Hogwild algorithms as a framewo
 rk. Our implementations on this novel architecture achieved superior hardw
 are efficiency and scalability over that of a conventional cluster using M
 PI. Furthermore these improvements were gained using naive data partitioni
 ng techniques and hardware with substantially less compute capability than
  that present in conventional systems.\n\nTag: Online Only, Algorithms, Ex
 treme Scale Computing\n\nRegistration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR