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:20220812T074358Z
LOCATION:Singapore Room
DTSTART;TZID=Europe/Stockholm:20220629T140000
DTEND;TZID=Europe/Stockholm:20220629T160000
UID:submissions.pasc-conference.org_PASC22_sess155@linklings.com
SUMMARY:MS6D - High-Performance Machine Learning: Scale and Performance (P
 art II)
DESCRIPTION:Minisymposium\n\nWith the growing range of machine learning ap
 proaches and the availability of large-scale data, performant and scalable
  algorithmic techniques have become central topics. Nowadays, high-perform
 ance computing (HPC) technologies have become essential for modern large-s
 cale machine learning applications. The convergence of HPC and machine lea
 rning has shown compelling success across many application fields. With th
 at said, real-world applications are rife with inherent challenges associa
 ted with high dimensionality, scarcity of data, and ill-posedness, among m
 any others. This minisymposium will serve as a platform to discuss cutting
 -edge developments for modern, scalable, and efficient machine learning ap
 proaches.\n\nScaling Distributed Training of Regular and Irregular Deep Le
 arning Models on Public Cloud\n\nKarypis\n\nLearning from graph and relati
 onal data plays a major role in many applications including social network
  analysis, marketing, e-commerce, information retrieval, knowledge modelin
 g, medical and biological sciences, engineering, and others. In the last f
 ew years, Graph Neural Networks (GNNs) have emer...\n\n-------------------
 --\nNeural-Network Approaches for High-Dimensional Optimal Control Problem
 s\n\nRuthotto\n\nWe consider neural network approaches to solving high-dim
 ensional optimal control problems with deterministic and randomly perturbe
 d dynamics. The training process simultaneously approximates the value fun
 ction of the control problem and identifies the part of the state space li
 kely to be visited by...\n\n---------------------\nMG-GCN: Scalable Multi-
 GPU GCN Training Framework\n\nBalin, Sancak, Catalyurek\n\nFull batch trai
 ning of Graph Convolutional Network (GCN) modelsis not feasible on a singl
 e GPU for large graphs containing tens of millionsof vertices or more. Rec
 ent work has shown that, for the graphs used in themachine learning commun
 ity, communication becomes a bottleneck and scalingis blocked o...\n\n----
 -----------------\nMachine Learning at Scale using ALP/GraphBLAS\n\nYzelma
 n\n\nWith Algebraic Programming (ALP), primitives require explicit algebra
 ic structures given by programmers when given data-centric operations on c
 ontainers. When considering linear algebraic relations and the duality bet
 ween graphs and sparse matrices, GraphBLAS emerges as an algebraic program
 ming mode...\n\n\nDomain: Computer Science and Applied Mathematics
END:VEVENT
END:VCALENDAR