Scientific Machine Learning: Our Own Experience

Abstract :

Scientific Machine Learning (SciML) has emerged as a discipline combining Machine Learning, Data-Science and Computational Methods leading to new powerful tools in the realm of Computational Simulation.
Here, I briefly describe our ower experience over the last ten years in combining Computational simulation, High-Performance computing, and Machine Learning (the core components of Scientific Machine Learning) to provide solutions in diGerent application fields. We particularly emphasize applications within the realm of Energy Transition, like, for instance, Seismic, CCS, and Wind Energy.

I go into more technical detail in two specific topics: Uncertainty Quantification in Carbone Capture and Sequestration computational modeling and (if time allows) Full Wave Inversion with Quantified Uncertainty.

Le mardi 24 septembre 2024 à 11h00 / Amphithéâtre François Canac, LMA

Fernando Rochinha, professor of mechanics at Universidade Federal do Rio de Janeiro (Brazil), will also give an introductory lecture on machine learning and artificial intelligence. Topics will include:

- Introduction and Context: Scientific Machine Learning (SciML)
- Predictive Models and Computational Simulation (Digital Twins)
- Machine Learning Models: Neural Networks (Deep Learning)
- Data Driven and Hybrid (Physics Aware Machine learning) Models
- Advanced topics: Probabilistic Learning (generative models)

This course will take place over 6 sessions of 2 hours, on September 17-18-19-24-25-26 from 2pm to 4pm. The course will take place on the LMA premises. 

Voir en ligne : plus d’informations concernant l’orateur