BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20200129T163559Z
LOCATION:502-503-504
DTSTART;TZID=America/Denver:20191118T140000
DTEND;TZID=America/Denver:20191118T143000
UID:submissions.supercomputing.org_SC19_sess115_ws_mlhpce107@linklings.com
SUMMARY:Parallel Data-Local Training for Optimizing Word2Vec Embeddings fo
 r Word and Graph Embeddings
DESCRIPTION:Workshop\n\nParallel Data-Local Training for Optimizing Word2V
 ec Embeddings for Word and Graph Embeddings\n\nMoon, Newman-Griffis, Kim, 
 Sukumaran-Rajam, Fosler-Lussier...\n\nThe Word2Vec model is a neural netwo
 rk-based unsupervised word embedding technique widely used in applications
  such as natural language processing, bioinformatics and graph mining. As 
 Word2Vec repeatedly performs Stochastic Gradient Descent (SGD) to minimize
  the objective function, it is very compute-intensive. However, existing m
 ethods for parallelizing Word2Vec are not optimized enough for data locali
 ty to achieve high performance. In this paper, we develop a parallel data-
 locality-enhanced Word2Vec algorithm based on Skip-gram with a novel negat
 ive sampling method that decouples loss calculation with positive and nega
 tive samples; this allows us to efficiently reformulate matrix-matrix oper
 ations for the negative samples over the sentence. Experimental results de
 monstrate our parallel implementations on multi-core CPUs and GPUs achieve
  significant performance improvement over the existing state-of-the-art pa
 rallel Word2Vec implementations while maintaining evaluation quality. We a
 lso show the utility of our Word2Vec implementation within the Node2Vec al
 gorithm which accelerates embedding learning for large graphs.\n\nTag: Wor
 kshop Reg Pass, Machine Learning\n\nRegistration Category: Workshop Reg Pa
 ss, Machine Learning
URL:https://sc19.supercomputing.org/presentation/?id=ws_mlhpce107&sess=ses
 s115
END:VEVENT
END:VCALENDAR