G-StAI

Description of the workshop

G-StAI is a 4-day workshop aimed at researchers working in geometric statistics and geometric methods for deep learning.

The workshop will be held at Metochi Study Center, Lesbos, Greece. Accomodation at the study center and all meals will be covered for all participants. For more images from center, see Metochi's Facebook page.

The workshop is May 6 to 9, but you are invited to stay the full week from May 4 to May 11, for collaboration or to have time to focus on your writing. The center is localed in a former Greek monastery and allows for a relaxing atmosphere for you to collaborate or be alone with your thoughts.

Participation by invite only.

Our aim with the workshop is to bring together experts on differential geometry, geometric statistics and deep learning to spawn discussion across these fields. In several important areas, using geometric ideas in machine learning and deep learning has lately proved extremely useful, for example for deep learning on Lie groups and manifolds, for equivariant networks, and for latent representations with non-Euclidean spaces. Conversely, machine learning gives new applications and insight in geometry with examples being the theoretical interest in optimal transport and relaxed versions of this, and the use of neural networks to learn stochastic dynamics and scores on geometric spaces.

With the workshop, we have the possibility to bring top experts from geometry, geometric statistics and machine learning together in a small workshop format setting with informal discussions that we hope can lead to new ideas, new collaborations and new results. We very much hope you will join us in this.