Research Foundations
Geometry, Probability & the Soul of Every Model. "If you want to understand the soul of a model, you must speak the language of Geometry and Probability.".
Information Geometry & Natural Gradients - Master the Fisher Information Matrix as a Riemannian metric Optimal Transport & Wasserstein Geometry - Understand the Monge-Kantorovich problem Differential Geometry & Geometric Deep Learning - Master Equivariant Neural Networks (CNNs as a special case)
The Equivariant Transformer.
