Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2508.10531 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917421353271296 |
|---|---|
| author | Luan, Hao Goh, Yi Xian Ng, See-Kiong Ling, Chun Kai |
| author_facet | Luan, Hao Goh, Yi Xian Ng, See-Kiong Ling, Chun Kai |
| contents | Modifications to test-time sampling have emerged as an important extension to diffusion algorithms, with the goal of biasing the generative process to achieve a given objective without having to retrain the entire diffusion model. However, generating jointly correlated samples from multiple pre-trained diffusion models while simultaneously enforcing task-specific constraints without costly retraining has remained challenging. To this end, we propose Projected Coupled Diffusion (PCD), a novel test-time framework for constrained joint generation. PCD introduces a coupled guidance term into the generative dynamics to encourage coordination between diffusion models and incorporates a projection step at each diffusion step to enforce hard constraints. Empirically, we demonstrate the effectiveness of PCD in application scenarios of image-pair generation, object manipulation, and multi-robot motion planning. Our results show improved coupling effects and guaranteed constraint satisfaction without incurring excessive computational costs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_10531 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Projected Coupled Diffusion for Test-Time Constrained Joint Generation Luan, Hao Goh, Yi Xian Ng, See-Kiong Ling, Chun Kai Machine Learning Modifications to test-time sampling have emerged as an important extension to diffusion algorithms, with the goal of biasing the generative process to achieve a given objective without having to retrain the entire diffusion model. However, generating jointly correlated samples from multiple pre-trained diffusion models while simultaneously enforcing task-specific constraints without costly retraining has remained challenging. To this end, we propose Projected Coupled Diffusion (PCD), a novel test-time framework for constrained joint generation. PCD introduces a coupled guidance term into the generative dynamics to encourage coordination between diffusion models and incorporates a projection step at each diffusion step to enforce hard constraints. Empirically, we demonstrate the effectiveness of PCD in application scenarios of image-pair generation, object manipulation, and multi-robot motion planning. Our results show improved coupling effects and guaranteed constraint satisfaction without incurring excessive computational costs. |
| title | Projected Coupled Diffusion for Test-Time Constrained Joint Generation |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2508.10531 |