VAE toolbox

VAE toolbox is a project aimed at making recently published deep learning methods for identifying targeted teleconnections and circulation regimes usable to the scientific community. For this purpose, we develop an open-source Python package containing the variational autoencoder (VAE)-based methods in a modular and accessible way. Additionally, we apply and compare these methods in an energy meteorology case study showcasing the advantages of the VAE-based methods.

Mentors

  • Edward Comyn-Platt
  • Clara Ducher
  • Chiara Cagnazzo
  • Fiona Spuler
  • Ben Aslan

Participants

Alexander Hempel

Antonia Bahr