Joint Seminar: Rise of the machines: deep learning methods in oceanography
In many fields the steady rise in computing resources and the amount of data has made it possible to learn about
our environment without considering first principles or laws of
nature. In physical oceanography, however, that is not obviously
true. Partly this is due to a paucity of data, but also because
the numerical tools are rather brittle, and the laws of physics
are powerful and have been tested for centuries. Still, progress
can be made and I will discuss four areas where higher-level
statistics led to new insights: Bayesian optimization of GCMs,
likelihood of rogue waves, equation discovery for the AMOC, and
causal inference for the glacial carbon cycle.
Datum
09.07.2025
Uhrzeit
13:30 h
Ort
- Bundesstr. 53, room 022/023
- Seminar Room 022/023, Ground Floor, Bundesstrasse 53, 20146 Hamburg, Hamburg
Chair
- Johann Jungclaus