BEGIN:VCALENDAR
VERSION:2.0
PRODID:https://www.uni-due.de
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:Europe/Berlin
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700329T020000
RRULE:FREQ=YEARLY;INTERVAL=1;BYDAY=-1SU;BYMONTH=3
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701025T030000
RRULE:FREQ=YEARLY;INTERVAL=1;BYDAY=-1SU;BYMONTH=10
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:ude20241216160000
CLASS:PUBLIC
SUMMARY:Seminar Numerische Mathematik und Mechanik, "Neural Networks"
DTSTART;TZID=Europe/Berlin:20241216T160000
DTEND;TZID=Europe/Berlin:20241216T173000
DTSTAMP:20241216T160000Z
LOCATION;ENCODING=QUOTED-PRINTABLE:Campus Campus Essen : R12 S02 H85
CONTACT:Herr Dr.-Ing. Maximilian Reichel (Institut für Mechanik)
DESCRIPTION:Herr Dr.-Ing. Maximilian Reichel (Institut für Mechanik)
Seminar Numerische Mathematik und Mechanik, "Neural Networks"
Dr. Evelyn Herberg, Ruprecht-Karls-University Heidelberg, Interdisciplinary Center for Scientific Computing (IWR)
In this talk we introduce Neural Networks from a mathematical perspective and discuss challenges, especially over- and underfitting as well as hyperparameter choice. Subsequently, several network architectures and their applications are presented. Here, we focus on architectures suited for solving differential equations and the various options to model the problem, in particular which data is required for training of the network. Finally, we explore recent research in adaptive pruning algorithms based on a specialized regularization technique and further research directions with potential application in engineering problems.

Interessenten sind herzlich eingeladen, es wird jedoch um kurze Anmeldung bei Herrn Reichel gebeten (maximilian.reichel@uni-due.de). 


Monday, 16. December 2024
END:VEVENT
END:VCALENDAR