Dott. Giacomo GRECO
Università di Roma "Tor Vergata"
Abstract
In this talk we will address the sampling problem, namely how we can effectively sample from an unknown distribution given a finite number of samples from it. In particular we will describe how common sampling strategies can be improved by considering score based diffusion models in generative AI. Lastly, we will adopt a Gamma and Malliavin Calculi point of view in order to generalize Score-based diffusion Generative Models (SGMs) to an infinite-dimensional abstract Hilbertian setting, by means of Dirichlet forms associated to the Cameron-Martin space of Gaussian measures and Wiener chaoses. Based on https://arxiv.org/abs/2505.13189
