It's Never Too Late: Noise Optimization for Collapse Recovery in Trained Diffusion Models
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_ | 1866909005997146112 |
|---|---|
| author | Harrington, Anne Koepke, A. Sophia Karthik, Shyamgopal Darrell, Trevor Efros, Alexei A. |
| author_facet | Harrington, Anne Koepke, A. Sophia Karthik, Shyamgopal Darrell, Trevor Efros, Alexei A. |
| contents | Contemporary text-to-image models exhibit a surprising degree of mode collapse, as can be seen when sampling several images given the same text prompt. Previous work has attempted to address this issue by steering the model using guidance mechanisms, or by generating a large pool of candidates and refining them. In this work, we take a different direction and aim for diversity in generations via noise optimization. Specifically, we show that a simple noise optimization objective can mitigate mode collapse while preserving the fidelity of the base model. We also analyze the frequency characteristics of the noise and show that alternative noise initializations with different frequency profiles can improve both optimization and search. Our experiments demonstrate that noise optimization yields superior results in terms of generation quality and diversity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_00090 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | It's Never Too Late: Noise Optimization for Collapse Recovery in Trained Diffusion Models Harrington, Anne Koepke, A. Sophia Karthik, Shyamgopal Darrell, Trevor Efros, Alexei A. Computer Vision and Pattern Recognition Machine Learning Contemporary text-to-image models exhibit a surprising degree of mode collapse, as can be seen when sampling several images given the same text prompt. Previous work has attempted to address this issue by steering the model using guidance mechanisms, or by generating a large pool of candidates and refining them. In this work, we take a different direction and aim for diversity in generations via noise optimization. Specifically, we show that a simple noise optimization objective can mitigate mode collapse while preserving the fidelity of the base model. We also analyze the frequency characteristics of the noise and show that alternative noise initializations with different frequency profiles can improve both optimization and search. Our experiments demonstrate that noise optimization yields superior results in terms of generation quality and diversity. |
| title | It's Never Too Late: Noise Optimization for Collapse Recovery in Trained Diffusion Models |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2601.00090 |