Inverse Bridge Matching Distillation

2025-02-03

AI Summary

Diffusion Bridge Models (DBMs) suffer from slow inference and poor deployability in image translation tasks. To address this, we propose a novel distillation framework based on inverse bridge matching, enabling the first unified single-step compression of both conditional and unconditional DBMs. Our method requires only noisy-image training data—no clean ground-truth supervision—and theoretically yields a closed-form, analytically tractable objective function. It is compatible with various Score Matching variants and one-step denoising network architectures. Extensive experiments on super-resolution, JPEG artifact removal, and sketch generation demonstrate 4×–100× inference speedup over teacher DBMs. Remarkably, under certain configurations, our distilled models even surpass the teacher in FID and LPIPS metrics, achieving unprecedented efficiency-quality trade-offs.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Data Compression

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Learning diffusion bridge models is easy; making them fast and practical is an art. Diffusion bridge models (DBMs) are a promising extension of diffusion models for applications in image-to-image translation. However, like many modern diffusion and flow models, DBMs suffer from the problem of slow inference. To address it, we propose a novel distillation technique based on the inverse bridge matching formulation and derive the tractable objective to solve it in practice. Unlike previously developed DBM distillation techniques, the proposed method can distill both conditional and unconditional types of DBMs, distill models in a one-step generator, and use only the corrupted images for training. We evaluate our approach for both conditional and unconditional types of bridge matching on a wide set of setups, including super-resolution, JPEG restoration, sketch-to-image, and other tasks, and show that our distillation technique allows us to accelerate the inference of DBMs from 4x to 100x and even provide better generation quality than used teacher model depending on particular setup.

Problem

• Diffusion Bridge Models
• Image Translation
• Processing Speed

Innovation

• Reversed Connection Matching Refinement
• Diffusion Bridge Models
• Image Transformation Acceleration