BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Accelerating HPC Applications Using Machine Learning-based Sur
 rogates
X-WR-TIMEZONE:Pacific Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20260715T231948Z
UID:tag:localist.com\,2008:EventInstance_49462852904384
DTSTART:20250425T210000Z
DTEND:20250425T220000Z
DESCRIPTION:Watch the recording here: https://media.oregonstate.edu/media/t
 /1_ny89iggj/232665543\n\n \n\nZoom Link: https://oregonstate.zoom.us/s/983
 57211915\n\n \n\nTalk Info: Large-scale scientific simulations drive scien
 tific and engineering discovery across many domains\, but face performance
  problems. These simulations typically involve complex physics computation
 s but are difficult to scale efficiently on high-performance hardware. Mac
 hine Learning (ML)-based surrogates are revolutionizing the natural scienc
 es. By employing reverse-engineering and automatic learning methodologies\
 , ML surrogates can solve complex\, unstructured problems with less comput
 ing power and execution time than needed by traditional direct and first-p
 rinciple methods\, with the advantages of lower implementation costs\, str
 onger generalization capability\, and lower computation overheads. In this
  talk\, I will introduce the application of ML-based surrogates in two use
 r-inspired High-Performance Computing (HPC) applications\, and a generaliz
 ed framework that automates the application process of ML-based surrogates
  in the HPC domain. Auto-HPCnet democratizes the usage of ML-based surroga
 tes and is the first end-to-end framework that makes past proposals for th
 e ML-based surrogate model practical and disciplined. Auto-HPCnet introduc
 es a workflow to address unique challenges when applying the surrogates\, 
 such as feature acquisition and meeting the application-specific constrain
 t on the quality of the final computation outcome. Also\, the two real-wor
 ld HPC applications include the Eulerian fluid simulation and the AC-OPF p
 ower grid simulation. In the Eulerian fluid simulation\, we generate multi
 ple surrogate candidates before the simulation and introduce a runtime sch
 eduler that dynamically switches the ML surrogates to make the best effort
 s to reach the user’s requirement on simulation quality. We show that ou
 r method achieves 1.46x and 590x speedup\, compared with a state-of-the-ar
 t ML model and the numerical fluid simulation respectively\, while providi
 ng better simulation quality than the state-of-the-art model. In the AC-OP
 F power grid simulation\, we generate multitask-learning (MTL) surrogates 
 to predict the initial values of variables critical to the convergence of 
 the power grid problem. The MTL models allow information sharing when pred
 icting multiple dependent variables\, while including customized layers to
  predict individual variables. The MTL model incorporates physics-informed
  learning to improve model accuracy and interpretability. These techniques
  bring 2.60× speedup on average (up to 3.28×) computed over 10\,000 larg
 e-scale power grid problems\, without losing solution optimality.\n\n \n\n
 Bio: Wenqian Dong is an assistant professor in the EECS department at Oreg
 on State University. She earned her Ph.D. in EECS at the University of Cal
 ifornia\, Merced\, in Spring 2022. Recently\, she is selected for the IEEE
 -CS Technical Community on High Performance Computing (TCHPC) Early Career
  Researchers Award for Excellence in High Performance Computing. Her resea
 rch focuses on three main areas. She has contributed significantly to scie
 ntific machine learning\, particularly in using machine learning to speed 
 up HPC applications. Her work is showcased in conferences like SC’19 and
  SC’20. Wenqian has excelled in automatic machine learning\, concentrati
 ng on creating machine learning models for HPC applications. Her papers in
  VLDB’21\, HPDC’23\, and ASPLOS’22 highlight her notable contributio
 ns. She’s skilled in optimizing system performance\, aiming to enhance H
 PC applications’ quality and efficiency through system optimization. Her
  work presented at conferences like ICS’21\, Eurosys’21\, ICPP’18\, 
 and Parallel Computing’23 illustrates her dedication to this field. Her 
 work has generated real impacts in the HPC community. For example\, her wo
 rk on power grid simulation using ML leads to 3.28 times performance impro
 vement and highlighted Newswise as a DOE science innovation. Furthermore\,
  Wenqian is committed to enrich the HPC community. Her commitment is appar
 ent in her various roles as an organizer for ICPP’24\, the MLBench’23 
 workshop\, and as a member of the Technical Program Committee (TPC) for th
 e SC’24\, HPDC’24\, CCGrid’24\, IEEE Cloud 2023\, IEEE Cluster 2023\
 , AI4Science 2022 workshop\, and the GPGPU 2023 workshop.
GEO:44.565762;-123.281717
LOCATION:Learning Innovation Center (LINC)\, 302
SUMMARY:Accelerating HPC Applications Using Machine Learning-based Surrogat
 es
URL;VALUE=URI:https://events.oregonstate.edu/event/accelerating-hpc-applica
 tions-using-machine-learning-based-surrogates
CATEGORIES:Lecture or Presentation
END:VEVENT
END:VCALENDAR