Timezone: »

A Continuous Time Framework for Discrete Denoising Models
Andrew Campbell · Joe Benton · Valentin De Bortoli · Thomas Rainforth · George Deligiannidis · Arnaud Doucet

Thu Dec 01 09:00 AM -- 11:00 AM (PST) @ Hall J #909

We provide the first complete continuous time framework for denoising diffusion models of discrete data. This is achieved by formulating the forward noising process and corresponding reverse time generative process as Continuous Time Markov Chains (CTMCs). The model can be efficiently trained using a continuous time version of the ELBO. We simulate the high dimensional CTMC using techniques developed in chemical physics and exploit our continuous time framework to derive high performance samplers that we show can outperform discrete time methods for discrete data. The continuous time treatment also enables us to derive a novel theoretical result bounding the error between the generated sample distribution and the true data distribution.

Author Information

Andrew Campbell (University of Oxford)
Joe Benton (University of Oxford)
Joe Benton

PhD student in the Department of Statistics at the University of Oxford, supervised by Arnaud Doucet and George Deligiannidis. Working on developing a theoretical understanding of generative modelling techniques, with a particular focus on diffusion models and variational autoencoders.

Valentin De Bortoli (ENS Ulm, CNRS)
Thomas Rainforth (University of Oxford)
George Deligiannidis (Oxford)
Arnaud Doucet (Oxford)

More from the Same Authors