Resolving Endpoint Underfitting in Diffusion Bridges via Noise Alignment
Quick Take
The Noise-Aligned Diffusion Bridge (NADB) addresses endpoint underfitting in diffusion bridge models by correcting noise mismatches, enhancing performance in image restoration and translation tasks. This approach utilizes a mean network for cleaner targets and introduces a novel mapping relationship, demonstrating effectiveness through experimental validation.
Key Points
- NADB reformulates diffusion bridges to resolve significant drift in predicted variance.
- The approach uses a mean network for cleaner conditional targets.
- Experimental validation shows improved performance in multiple image tasks.
- Noise alignment corrects underfitting near the target distribution endpoint.
- Code for NADB is available on GitHub.
Article Content
From source RSS / original summaryarXiv:2605. 28962v1 Announce Type: new Abstract: Diffusion bridge models offer a powerful framework for connecting two data distributions, such as in image restoration and translation. Many existing methods learn this bridge by mimicking the score-matching formulation of standard diffusion models. In this work, we find that this way leads to an anomalous underfitting phenomenon near the target endpoint, as the process approaches the target distribution ($t \to 0$).
This underfitting, characterized by significant drift in the predicted variance and direction, results from an excessively large discrepancy in noise levels between the network's input and its regression target. To resolve this issue, we propose the Noise-Aligned Diffusion Bridge (NADB). Our approach reformulates the diffusion bridge by first employing a mean network to provide a cleaner conditional target, and then introducing a novel, noise-aligned mapping relationship.
This new formulation resolves the noise mismatch and corrects the underfitting near the target endpoint. Experimental validation across multiple image restoration and image translation tasks demonstrates the effectiveness of our approach. Code is available at https://github. com/gyr02/NADB.
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →Evi-Steer: Learning to Steer Biomedical Vision-Language Models through Efficient and Generalizable Evidential Tuning
Evi-Steer introduces a novel evidential tuning framework for BiomedCLIP, achieving 0.11% parameter updates while enhancing uncertainty-aware fine-tuning. It outperforms state-of-the-art methods across 15 biomedical imaging datasets, proving effective in few-shot learning and domain shifts for clinical applications.