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:20260522T150119Z
LOCATION:C140/142
DTSTART;TZID=America/Chicago:20181115T133000
DTEND;TZID=America/Chicago:20181115T150000
UID:submissions.supercomputing.org_SC18_sess190@linklings.com
SUMMARY:Deep Learning
DESCRIPTION:Exploring Flexible Communications for Streamlining DNN Ensembl
 e Training Pipelines\n\nParallel training of a Deep Neural Network (DNN) e
 nsemble on a cluster of nodes is a common practice to train multiple model
 s in order to construct a model with a higher prediction accuracy. Existin
 g ensemble training pipelines can perform a great deal of redundant operat
 ions, resulting in unnecessa...\n\n\nRandall Pittman, Hui Guan, and Xipeng
  Shen (North Carolina State University) and Seung-Hwan Lim and Robert M. P
 atton (Oak Ridge National Laboratory)\n---------------------\nAnatomy of H
 igh-Performance Deep Learning Convolutions on SIMD Architectures\n\nConvol
 ution layers are prevalent in many classes of deep neural networks, includ
 ing Convolutional Neural Networks (CNNs) which provide state-of-the-art re
 sults for tasks like image recognition, neural machine translation, and sp
 eech recognition. The computationally expensive nature of a convolution ..
 .\n\n\nEvangelos Georganas, Sasikanth Avancha, Kunal Banerjee, Dhiraj Kala
 mkar, Greg Henry, Hans Pabst, and Alexander Heinecke (Intel Corporation)\n
 ---------------------\nCosmoFlow: Using Deep Learning to Learn the Univers
 e at Scale\n\nDeep learning is a promising tool to determine the physical 
 model that describes our universe.   To handle the considerable computatio
 nal cost of this problem, we present CosmoFlow: a highly scalable deep lea
 rning application built on top of the TensorFlow framework.\n\nCosmoFlow u
 ses efficient implem...\n\n\nAmrita Mathuriya (Intel Corporation); Deborah
  Bard (National Energy Research Scientific Computing Center (NERSC), Lawre
 nce Berkeley National Laboratory); Pete Mendygral (Cray Inc); Lawrence Mea
 dows (Intel Corporation); James Arnemann (University of California, Berkel
 ey); Lei Shao (Intel Corporation); Siyu He (Carnegie Mellon University); T
 uomas Karna (Intel Corporation); Diana Moise (Cray Inc); Simon J. Pennycoo
 k (Intel Corporation); Kristyn Maschhoff (Cray Inc); Jason Sewall and Nali
 ni Kumar (Intel Corporation); Shirley Ho (Lawrence Berkeley National Labor
 atory, Carnegie Mellon University); Michael F. Ringenburg (Cray Inc); Mr P
 rabhat (Lawrence Berkeley National Laboratory, National Energy Research Sc
 ientific Computing Center (NERSC)); and Victor Lee (Intel Corporation)\n\n
 Tag: Applications, Cosmology, Data Analytics, Deep Learning, Machine Learn
 ing, Programming Systems, Storage, Visualization\n\nRegistration Category:
  Tech Program Reg Pass\n\nSession Chair: Tal Ben-Nun (Lawrence Livermore N
 ational Laboratory (LLNL))
END:VEVENT
END:VCALENDAR