Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.08144 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915995033010176 |
|---|---|
| author | Wu, Fang Zhao, Haokai Xing, Da Cao, Hanqun Xu, Tinson Li, Yanchao Tang, Xiangru Wang, Zehong Tu, Aaron Pang, Kuan Wang, Hanchen Lin, Hongbin Zhou, Zeqi Li, Yinxi Xia, Peng Li, Li Erran Tao, Molei Leskovec, Jure Joshi, Aditya Choi, Yejin |
| author_facet | Wu, Fang Zhao, Haokai Xing, Da Cao, Hanqun Xu, Tinson Li, Yanchao Tang, Xiangru Wang, Zehong Tu, Aaron Pang, Kuan Wang, Hanchen Lin, Hongbin Zhou, Zeqi Li, Yinxi Xia, Peng Li, Li Erran Tao, Molei Leskovec, Jure Joshi, Aditya Choi, Yejin |
| contents | Diffusion models have achieved remarkable success across a wide range of generative tasks, yet their training paradigm largely treats injected noise as uniformly informative. In this work, we challenge this assumption and introduce NoiseRater, a meta-learning framework for instance-level noise valuation in diffusion model training. We propose a parametric noise rater that assigns importance scores to individual noise realizations conditioned on data and timestep, enabling adaptive reweighting of the training objective. The rater is trained via bilevel optimization to improve downstream validation performance after inner-loop diffusion updates. To enable efficient deployment, we further design a decoupled two-stage pipeline that transitions from soft weighting during meta-training to hard noise selection during standard training. Extensive experiments on FFHQ and ImageNet demonstrate that not all noise samples contribute equally, and that prioritizing informative noise improves both training efficiency and generation quality. Our results establish noise valuation as a complementary and previously underexplored axis for improving diffusion model training. Our code is available at: https://anonymous.4open.science/r/NoiseRater-DEB116. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_08144 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | NoiseRater: Meta-Learned Noise Valuation for Diffusion Model Training Wu, Fang Zhao, Haokai Xing, Da Cao, Hanqun Xu, Tinson Li, Yanchao Tang, Xiangru Wang, Zehong Tu, Aaron Pang, Kuan Wang, Hanchen Lin, Hongbin Zhou, Zeqi Li, Yinxi Xia, Peng Li, Li Erran Tao, Molei Leskovec, Jure Joshi, Aditya Choi, Yejin Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition Diffusion models have achieved remarkable success across a wide range of generative tasks, yet their training paradigm largely treats injected noise as uniformly informative. In this work, we challenge this assumption and introduce NoiseRater, a meta-learning framework for instance-level noise valuation in diffusion model training. We propose a parametric noise rater that assigns importance scores to individual noise realizations conditioned on data and timestep, enabling adaptive reweighting of the training objective. The rater is trained via bilevel optimization to improve downstream validation performance after inner-loop diffusion updates. To enable efficient deployment, we further design a decoupled two-stage pipeline that transitions from soft weighting during meta-training to hard noise selection during standard training. Extensive experiments on FFHQ and ImageNet demonstrate that not all noise samples contribute equally, and that prioritizing informative noise improves both training efficiency and generation quality. Our results establish noise valuation as a complementary and previously underexplored axis for improving diffusion model training. Our code is available at: https://anonymous.4open.science/r/NoiseRater-DEB116. |
| title | NoiseRater: Meta-Learned Noise Valuation for Diffusion Model Training |
| topic | Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.08144 |