BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241120T082409Z
LOCATION:HG D 1.2
DTSTART;TZID=Europe/Stockholm:20240604T170000
DTEND;TZID=Europe/Stockholm:20240604T173000
UID:submissions.pasc-conference.org_PASC24_sess124_msa283@linklings.com
SUMMARY:Algebraic Programming for Graph & Machine Learning
DESCRIPTION:Minisymposium\n\nAlbert-Jan Yzelman, Aristeidis Mastoras, and 
 Alberto Scolari (Huawei)\n\nAlgebraic Programming, or ALP for short, allow
 s for programming with explicit algebraic structures. Such structures rang
 e from the simplistic such as associative binary operators, to richer cons
 tructs such as semirings. This algebraic knowledge, given to ALP by the pr
 ogrammer, percolates through the ALP framework and allows it to auto-paral
 lelise, as well as perform other types of automatic program transformation
 s for achieving high performance.\nOriginating from the GraphBLAS, ALP’s i
 nitial and most mature interface concerns generalized linear algebra. This
  talk focuses on two recent works related to the use of ALP/GraphBLAS with
 in high-performance machine learning. First, we shall overview how the ALP
  framework achieves interoperability with Spark, making accessible ALP/Gra
 phBLAS algorithms from within Spark, a standard framework for Big Data ana
 lytics. Second, the talk details the ALP/Pregel implementation, which reli
 es on ALP/GraphBLAS to simulate, at high efficiency and at full scalabilit
 y, vertex-centric programming.\nPerformance comparisons of canonical machi
 ne learning algorithms such as PageRank versus the standard GraphX library
  on top of Spark display significant performance gains of up to 21x on two
  nodes. The vertex-centric ALP/Pregel furthermore is shown to auto-paralle
 lise well on shared-memory architectures; a vertex-centric PageRank-like a
 lgorithm achieves a 5.7x speedup versus a highly optimised parallel linear
  algebraic PageRank.\n\nDomain: Computational Methods and Applied Mathemat
 ics\n\nSession Chairs: Dimosthenis Pasadakis (Università della Svizzera it
 aliana) and Olaf Schenk (Università della Svizzera italiana, ETH Zurich)
END:VEVENT
END:VCALENDAR