We introduce Functional Diffusion Processes (FDPs), which generalize score-based diffusion models to infinite-dimensional function spaces. FDPs require a new mathematical framework to describe the forward and backward dynamics, and several extensions to derive practical training objectives. These include infinite-dimensional versions of Girsanov theorem, in order to be able to compute an ELBO, and of the sampling theorem, in order to guarantee that functional evaluations in a countable set of points are equivalent to infinite-dimensional functions. We use FDPs to build a new breed of generative models in function spaces, which do not require specialized network architectures, and that can work with any kind of continuous data.Our results on real data show that FDPs achieve high-quality image generation, using a simple MLP architecture with orders of magnitude fewer parameters than existing diffusion models.
Continuous-time functional diffusion processes
NeurIPS 2023, 37th Conference on Neural Information Processing Systems, 11-16 December 2023, New Orleans, USA
© NIST. Personal use of this material is permitted. The definitive version of this paper was published in NeurIPS 2023, 37th Conference on Neural Information Processing Systems, 11-16 December 2023, New Orleans, USA and is available at :
PERMALINK : https://www.eurecom.fr/publication/7229