BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Can AI weather and climate emulators predict out-of-distribution g
 ray swan extreme events? - Prof Pedram Hassanzedeh\, University of Chicago
DTSTART:20250516T150000Z
DTEND:20250516T160000Z
UID:TALK231574@talks.cam.ac.uk
CONTACT:Professor Grae Worster
DESCRIPTION:Artificial intelligence (AI) is transforming weather and clima
 te modeling. For example\, neural network-based weather models can now out
 perform physics-based models for up to 15-day forecasts at a fraction of t
 he computing time. However\, these AI models have challenges with learning
  the rarest yet most impactful weather extremes\, particularly the gray sw
 ans (i.e.\, physically possible events so rare they have never been seen i
 n the training set). They also poorly learn multi-scale chaotic dynamics. 
 I will discuss some of these challenges\, as well as some of the surprisin
 g capabilities of these models\, e.g.\, transferring what they learn from 
 one region to another for dynamically similar event. I will present ideas 
 around integrating tools from applied math\, climate physics\, and AI to a
 ddress some of these challenges and make progress. In particular\, I will 
 discuss the use if rare event sampling algorithms and the Fourier transfor
 m and adjoint of the deep neural networks. 
LOCATION:MR2
END:VEVENT
END:VCALENDAR