Adversarial Schrödinger Bridge Matching

Nikita Gushchin,Daniil Selikhanovych,Sergei Kholkin,Evgeny Burnaev,Alexander Korotin

2024-05-23

Abstract:The Schrödinger Bridge (SB) problem offers a powerful framework for combining optimal transport and diffusion models. A promising recent approach to solve the SB problem is the Iterative Markovian Fitting (IMF) procedure, which alternates between Markovian and reciprocal projections of continuous-time stochastic processes. However, the model built by the IMF procedure has a long inference time due to using many steps of numerical solvers for stochastic differential equations. To address this limitation, we propose a novel Discrete-time IMF (D-IMF) procedure in which learning of stochastic processes is replaced by learning just a few transition probabilities in discrete time. Its great advantage is that in practice it can be naturally implemented using the Denoising Diffusion GAN (DD-GAN), an already well-established adversarial generative modeling technique. We show that our D-IMF procedure can provide the same quality of unpaired domain translation as the IMF, using only several generation steps instead of hundreds.

Machine Learning

What problem does this paper attempt to address?

This paper proposes a new approach to solve the Schrödinger Bridge problem within the combined framework of optimal transport and diffusion models. The traditional Iterative Markov Fitting (IMF) method for solving the Schrödinger Bridge problem requires a large number of numerical solution steps for stochastic differential equations, resulting in lengthy inference time. To address this issue, the paper introduces the Discrete-time IMF (D-IMF) method, which simplifies the inference process by learning a small number of transition probabilities in discrete time instead of continuous-time stochastic process learning. In practice, D-IMF can achieve the same quality as IMF with only a few generation steps, whereas IMF typically requires hundreds of steps. Specifically, the D-IMF method leverages adversarial generative network techniques, particularly Denoising Diffusion GAN (DD-GAN), to achieve discrete Markov projection. The paper also presents convergence speed analysis and closed-form update formulas for D-IMF in the context of high-dimensional Gaussian distributions, further consolidating its theoretical foundation. In the experimental section, the paper demonstrates the effectiveness of the D-IMF method through an image-to-image unpaired translation task. Compared to IMF, which requires hundreds of steps, D-IMF achieves similar results in just a few steps. Additionally, D-IMF exhibits advantages in tasks such as high-resolution text-to-image synthesis, indicating its potential in practical applications.

Adversarial Schrödinger Bridge Matching

Diffusion & Adversarial Schrödinger Bridges via Iterative Proportional Markovian Fitting

Diffusion Schrödinger Bridge Matching

Diffusion Bridge Mixture Transports, Schrödinger Bridge Problems and Generative Modeling

Diffusion Schrödinger Bridge with Applications to Score-Based Generative Modeling

Discrete Diffusion Schrödinger Bridge Matching for Graph Transformation

BM$^2$: Coupled Schrödinger Bridge Matching

Schrödinger Bridge Flow for Unpaired Data Translation

Provably Convergent Schrödinger Bridge with Applications to Probabilistic Time Series Imputation

Generalized Schrödinger Bridge Matching

Light and Optimal Schrödinger Bridge Matching

Deep Generative Learning via Schrödinger Bridge

Feedback Schrödinger Bridge Matching

Conditional Simulation Using Diffusion Schrödinger Bridges

Deep Generative Learning Via Schrodinger Bridge

Transport with Support: Data-Conditional Diffusion Bridges

Feedback Schr{ö}dinger Bridge Matching

Reflected Schrödinger Bridge for Constrained Generative Modeling

Diffusion Bridge Implicit Models

Generative modeling for time series via Schr{ö}dinger bridge

Stable generative modeling using Schrödinger bridges