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:20200129T163556Z
LOCATION:506
DTSTART;TZID=America/Denver:20191118T141000
DTEND;TZID=America/Denver:20191118T143500
UID:submissions.supercomputing.org_SC19_sess137_ws_xloop101@linklings.com
SUMMARY:Computational Strategies to Increase Efficiency of Gaussian-Proces
 s-Driven Autonomous Experiments
DESCRIPTION:Workshop\n\nComputational Strategies to Increase Efficiency of
  Gaussian-Process-Driven Autonomous Experiments\n\nNoack, Zwart\n\nNatural
  sciences are increasingly reliant on large volumes of experimental data o
 btained from highly-automated scientific equipment to drive scientific dis
 covery. While the instrumentation available to the scientific community en
 ables individual researchers to acquire experimental data at spectacular r
 ates, the decision making and experimental design process of the measureme
 nts is still largely driven by the individual researcher operating the ins
 trument.   \n\nIn recent years, some scientific areas, like X-Ray scatteri
 ng and synchrotron infra-red microscopy, have started to shift towards aut
 onomously executed experimentation, using tools from Artificial Intelligen
 ce and Machine Learning. \nGaussian Process Regression (GPR) is a popular 
 technique that enables the construction of a surrogate model required to d
 rive an experiment in an autonomous fashion. GPRs have been successfully i
 mplemented at synchrotron radiation beam lines at the Advanced Light Sourc
 e (ALS) and the National Synchrotron Light Source II (NSLS-II). \n\nGiven 
 that a single measurement on state-of-the-art synchrotron equipment can be
  acquired in a fraction of a second, being able to obtain feedback at a si
 milar time scale is essential. Traditionally, specific tasks in GPR, such 
 as hyper-parameter tuning and covariance estimation, are compute-intensive
 , limiting the utility of GPR as an appropriate tool when near real-time f
 eedback is an absolute necessity. In this paper we discuss, present and re
 view computational strategies that allow the numerical acceleration of Gau
 ssian Process analyses within a framework of autonomous sequential experim
 ents. The results show that significant time savings can be achieved by ta
 king advantage of a number of some rather well-established mathematical an
 d computational approaches.\n\nTag: Workshop Reg Pass, Experimental Datase
 ts, High-fidelity Simulations, Scalable Computing\n\nRegistration Category
 : Workshop Reg Pass, Experimental Datasets, High-fidelity Simulations, Sca
 lable Computing
URL:https://sc19.supercomputing.org/presentation/?id=ws_xloop101&sess=sess
 137
END:VEVENT
END:VCALENDAR