BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//jEvents 2.0 for Joomla//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:djer0iebpgabskk3n400te5umo@google.com
CATEGORIES:RTportal.ru
CREATED:20170606T092329
SUMMARY:Семинар "Структурные модели и глубинное обучение": Physics Informed Machine Learning
LOCATION:аудитория 615 ИППИ РАН
DESCRIPTION;ENCODING=QUOTED-PRINTABLE: Michael Chertkov, Adjunct Professor at Center for Energy Systems, SkolTech
 ; Los Alamos National Laboratory\n\nPhysics Informed Machine Learning\n\nMa
 chine Learning (statistical engineering) capabilities are in a phase of tre
 mendous growth. Underlying these advances is a strong and deep connection t
 o various aspects of statistical physics. There is also a great opportunity
  in pointing these tools toward physical modeling. In this presentation I i
 llustrate the two-way flow of ideas between physics and statistical enginee
 ring on three examples. First, I review the work on structure learning and 
 statistical estimation in power system distribution (thus physical) network
 s. Then I describe recent progress in constructive understanding of graph l
 earning (on example of inverse Ising model) illustrating that the generic i
 nverse task (of learning) is computationally easy in spite of the fact that
  the direct problem (inference or sampling) is difficult. I conclude specul
 ating how macro-scale models of physics (e.g. large eddy simulations of tur
 bulence) can be learned from micro-scale simulations (e.g. of Navier-Stocks
  equations). This event was imported from: https://rtportal.ru/index.php/ka
 lendar-sobytij/eventdetail/138/-/seminar-strukturnye-modeli-i-glubinnoe-obu
 chenie-physics-informed-machine-learning?tmpl=component
DTSTAMP:20260812T201258Z
DTSTART;TZID=Europe/Moscow:20160418T183000
DTEND;TZID=Europe/Moscow:20160418T200000
SEQUENCE:0
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR