Diffusion Bridge Implicit Models

TL;DR

This paper introduces diffusion bridge implicit models (DBIMs), a faster sampling method for denoising diffusion bridge models that maintains diversity and enables efficient image translation without extra training.

Contribution

It generalizes DDBMs with non-Markovian diffusion bridges, leading to significantly faster sampling and a new ODE formulation for improved numerical solutions.

Findings

DBIMs are up to 25 times faster than vanilla DDBMs.

They preserve generation diversity through booting noise.

The approach enables high-quality image translation and semantic interpolation.

Abstract

Denoising diffusion bridge models (DDBMs) are a powerful variant of diffusion models for interpolating between two arbitrary paired distributions given as endpoints. Despite their promising performance in tasks like image translation, DDBMs require a computationally intensive sampling process that involves the simulation of a (stochastic) differential equation through hundreds of network evaluations. In this work, we take the first step in fast sampling of DDBMs without extra training, motivated by the well-established recipes in diffusion models. We generalize DDBMs via a class of non-Markovian diffusion bridges defined on the discretized timesteps concerning sampling, which share the same marginal distributions and training objectives, give rise to generative processes ranging from stochastic to deterministic, and result in diffusion bridge implicit models (DBIMs). DBIMs are not only up to 25$\times$ faster than the vanilla sampler of DDBMs but also induce a novel, simple, and insightful form of ordinary differential equation (ODE) which inspires high-order numerical solvers. Moreover, DBIMs maintain the generation diversity in a distinguished way, by using a booting noise in the initial sampling step, which enables faithful encoding, reconstruction, and semantic interpolation in image translation tasks. Code is available at https://github.com/thu-ml/DiffusionBridge.