BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240626T180034Z
LOCATION:3001\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240625T171500
DTEND;TZID=America/Los_Angeles:20240625T173000
UID:dac_DAC 2024_sess105_RESEARCH1549@linklings.com
SUMMARY:SGM-PINN: Sampling Graphical Models for Faster Training of Physics
 -Informed Neural Networks
DESCRIPTION:Research Manuscript\n\nJohn Anticev, Ali Aghdaei, Wuxinlin Che
 ng, and Zhuo Feng (Stevens Institute of Technology)\n\nSGM-PINN is a graph
 -based importance sampling framework to improve the training efficacy of P
 hysics-Informed Neural Networks (PINNs) on parameterized problems. By appl
 ying a graph decomposition scheme to an undirected Probabilistic Graphical
  Model (PGM) built from the training dataset, our method generates node cl
 usters encoding conditional dependence between training samples. Biasing s
 ampling towards more important clusters allows smaller mini-batches and tr
 aining datasets, improving training speed and accuracy. We additionally fu
 se an efficient robustness metric with residual losses to determine region
 s requiring additional sampling. Experiments demonstrate the advantages of
  the proposed framework, achieving 3X faster convergence compared to prior
  state-of-the-art sampling methods.\n\nTopic: AI\n\nKeyword: AI/ML Applica
 tion and Infrastructure\n\nSession Chairs: Hongxiang Fan (Imperial College
  London; Samsung AI Center, UK) and Xiaoxuan Yang (University of Virginia,
  Stanford University)
END:VEVENT
END:VCALENDAR