Diffusion Classifier Guidance for Non-robust Classifiers
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912459534630912 |
|---|---|
| author | Vaeth, Philipp Kumar, Dibyanshu Paassen, Benjamin Gregorová, Magda |
| author_facet | Vaeth, Philipp Kumar, Dibyanshu Paassen, Benjamin Gregorová, Magda |
| contents | Classifier guidance is intended to steer a diffusion process such that a given classifier reliably recognizes the generated data point as a certain class. However, most classifier guidance approaches are restricted to robust classifiers, which were specifically trained on the noise of the diffusion forward process. We extend classifier guidance to work with general, non-robust, classifiers that were trained without noise. We analyze the sensitivity of both non-robust and robust classifiers to noise of the diffusion process on the standard CelebA data set, the specialized SportBalls data set and the high-dimensional real-world CelebA-HQ data set. Our findings reveal that non-robust classifiers exhibit significant accuracy degradation under noisy conditions, leading to unstable guidance gradients. To mitigate these issues, we propose a method that utilizes one-step denoised image predictions and implements stabilization techniques inspired by stochastic optimization methods, such as exponential moving averages. Experimental results demonstrate that our approach improves the stability of classifier guidance while maintaining sample diversity and visual quality. This work contributes to advancing conditional sampling techniques in generative models, enabling a broader range of classifiers to be used as guidance classifiers. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_00687 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Diffusion Classifier Guidance for Non-robust Classifiers Vaeth, Philipp Kumar, Dibyanshu Paassen, Benjamin Gregorová, Magda Machine Learning Computer Vision and Pattern Recognition Classifier guidance is intended to steer a diffusion process such that a given classifier reliably recognizes the generated data point as a certain class. However, most classifier guidance approaches are restricted to robust classifiers, which were specifically trained on the noise of the diffusion forward process. We extend classifier guidance to work with general, non-robust, classifiers that were trained without noise. We analyze the sensitivity of both non-robust and robust classifiers to noise of the diffusion process on the standard CelebA data set, the specialized SportBalls data set and the high-dimensional real-world CelebA-HQ data set. Our findings reveal that non-robust classifiers exhibit significant accuracy degradation under noisy conditions, leading to unstable guidance gradients. To mitigate these issues, we propose a method that utilizes one-step denoised image predictions and implements stabilization techniques inspired by stochastic optimization methods, such as exponential moving averages. Experimental results demonstrate that our approach improves the stability of classifier guidance while maintaining sample diversity and visual quality. This work contributes to advancing conditional sampling techniques in generative models, enabling a broader range of classifiers to be used as guidance classifiers. |
| title | Diffusion Classifier Guidance for Non-robust Classifiers |
| topic | Machine Learning Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.00687 |