Learning model parameters
28 Novembre 2024
ore: 17:00
AULA II - Piano Terra - Edificio in Bo (Via Francesco Selmi, 2 - Bologna)
Ingresso libero
-
44.4962966 11.3537311
By combining data assimilation and machine learning
Geophysical models have tunable parameters that are poorly known and manually chosen, causing errors. This talk explores objective parameter estimation methods, including stochastic neural networks, combining parameter estimation and data assimilation to reduce errors.