BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
BEGIN:STANDARD
DTSTART:19451014T230000
TZOFFSETFROM:+0630
TZOFFSETTO:+0530
TZNAME:IST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260630T105622Z
UID:2F083114-F238-4948-A923-6E7E1E1E2F54
DTSTART;TZID=Asia/Kolkata:20260701T180000
DTEND;TZID=Asia/Kolkata:20260701T200000
DESCRIPTION:A recent report by the US Department of Energy defines the area
  of scientific machine learning as\n“a core component of artificial inte
 lligence (AI) and a computational technology that can be trained\, with sc
 ientific data\, to augment or automate human skills”\, which has “the 
 potential to transform science and energy research”. In this presentatio
 n\, we discuss the potential of scientific machine learning methods to pro
 blems in computational electromagnetics starting from standard electromagn
 etic structure analysis and multi-physics modeling in the time-domain\, em
 ploying an unsupervised learning strategy based on Physics-Informed Neural
  Networks (PINN). PINNs directly integrate physical laws into their loss f
 unction\, so that the training process does not rely on the generation of 
 training ground truth data from simulations (as in typical neural networks
 ). We demonstrate the unconditionally stable solution of coupled electroma
 gnetic-thermal problems\, along with the modeling of frequency selective s
 urfaces and metasurfaces\, orders of magnitude faster than the conventiona
 l finite-difference time-domain technique (FDTD)\, including training time
  of the neural network.\n\nSpeaker(s): Costas Sarris\, \n\nVirtual: https:
 //events.vtools.ieee.org/m/565750
LOCATION:Virtual: https://events.vtools.ieee.org/m/565750
ORGANIZER:ayushi2023@iisc.ac.in
SEQUENCE:27
SUMMARY:Scientific machine learning for electromagnetic field computations
URL;VALUE=URI:https://events.vtools.ieee.org/m/565750
X-ALT-DESC:Description: <br /><p><span lang="EN-US" style="font-size: 11.0p
 t\; line-height: 107%\; font-family: 'Calibri'\,sans-serif\; mso-ascii-the
 me-font: minor-latin\; mso-fareast-font-family: Calibri\; mso-fareast-them
 e-font: minor-latin\; mso-hansi-theme-font: minor-latin\; mso-bidi-font-fa
 mily: 'Times New Roman'\; mso-bidi-theme-font: minor-bidi\; mso-ansi-langu
 age: EN-US\; mso-fareast-language: EN-US\; mso-bidi-language: AR-SA\;">A r
 ecent report by the US Department of Energy defines the area of scientific
  machine learning as <br>&ldquo\;a core component of artificial intelligen
 ce (AI) and a computational technology that can be trained\, with scientif
 ic data\, to augment or automate human skills&rdquo\;\, which has &ldquo\;
 the potential to transform science and energy research&rdquo\;. In this pr
 esentation\, we discuss the potential of scientific machine learning metho
 ds to problems in computational electromagnetics starting from standard el
 ectromagnetic structure analysis and multi-physics modeling in the time-do
 main\, employing an unsupervised learning strategy based on Physics-Inform
 ed Neural Networks (PINN). PINNs directly integrate physical laws into the
 ir loss function\, so that the training process does not rely on the gener
 ation of training ground truth data from simulations (as in typical neural
  networks). We demonstrate the unconditionally stable solution of coupled 
 electromagnetic-thermal problems\, along with the modeling of frequency se
 lective surfaces and metasurfaces\, orders of magnitude faster than the co
 nventional finite-difference time-domain technique (FDTD)\, <em>including 
 training time of the neural network.</em> </span></p>
END:VEVENT
END:VCALENDAR