Actualités IA

Vous trouverez dans cette page des informations concernant l'intelligence artificielle.

Workshops NVIDIA DLI (Deep Learning Institute)

Method Deep Learning - orienté sur le domaine de la prévision MTO (WF)

Publié le 25 avril 2025

Free DLI course: https://learn.nvidia.com/courses/course-detail?course_id=course-v1:DLI+S-FX-31+V1

This course explores the possibilities offered by state-of-the-art AI weather prediction models and teaches how to integrate them into custom workflows.

Main Goals:

  • Learn the fundamentals of AI weather simulation and understand the difference between AI-based and numerical weather prediction.
  • Gain hands-on experience running AI weather simulations with Earth2Studio for weather forecasting, historical analyses, and downscaling.
  • Get familiar with the individual parts of a weather inference pipeline: Data sources, perturbation, forecast and diagnostic models, and IO handling.
  • Validate the skill and calibration of weather forecasts with metrics commonly applied to deterministic and ensemble weather forecasts.
  • Build a custom AI weather model inference pipeline, using the example of coupling a global forecast model with a regional downscaling model.
  • Combine AI-generated weather data with downstream applications, following examples related to energy demand and production forecasts.