BEGIN:VCALENDAR
VERSION:2.0
CALSCALE:GREGORIAN
PRODID:UW-Madison-Physics-Events
BEGIN:VEVENT
SEQUENCE:1
UID:UW-Physics-Event-7981
DTSTART:20221214T170000Z
DTEND:20221214T181500Z
DTSTAMP:20260414T113549Z
LAST-MODIFIED:20221212T151944Z
LOCATION:Online Seminar: Please sign up for our mailing list at www.ph
 ysicsmeetsml.org for zoom link. We will also livestream the talk in Ch
 amberlin 5280.
SUMMARY:Differentiable Programming in HEP\, Physics ∩ ML Seminar\, L
 ukas Heinrich\, TU Munich
DESCRIPTION:The rise of machine learning within the last decade has to
  a large degree also been the success of differentiable programming an
 d gradient-based methods both in optimization as well as statistical i
 nference. Going beyond vanilla Deep Learning\, differentiable programm
 ing allows physicists to inject domain knowledge throughout the ML wor
 kflow from adding inductive bias to models via symmetries\, using phys
 ics models within the loss definition or in defining more informative 
 label data. While this may significantly increase both interpretabilit
 y and efficiency of ML application in physics\, challenges remain in c
 asting key physics processes in the language of differentiable program
 ming - particularly for the deeply hierarchical stochastic processes o
 ne observes in high energy physics. In this talk I will review recent 
 advances in applying differentiable programming as a paradigm to HEP a
 nd point out new research directions.
URL:https://www.physics.wisc.edu/events/?id=7981
END:VEVENT
END:VCALENDAR