Generative models have made significant advances in recent years, sparking an explosion of new applications with far-reaching societal implications. I will discuss the mathematical intuition behind diffusion models, the core technology behind recent art generation tools like DALL-E-2, Imagen, Stable Diffusion, Dreambooth, Lensa, and others. These applications introduce new technical challenges both for computational scalability and for controlling outputs to align results with users’ goals and ethical constraints. I will conclude by discussing an information-theoretic formulation of diffusion models that can be used to address these problems.

Dr. Greg Ver Steeg