Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.30553 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913171065798656 |
|---|---|
| author | Noël, Pierre-André |
| author_facet | Noël, Pierre-André |
| contents | I present diffusion models as part of a family of machine learning techniques that withhold information from a model's input and train it to guess the withheld information. I argue that diffusion's destroying approach to withholding is more flexible than typical hand-crafted information withholding techniques, providing a rich training playground that could be advantageous in some settings, notably data-scarce ones. I then address subtle issues that may arise when porting reinforcement learning techniques to the diffusion context, and wonder how such exploration problems could be addressed in more diffusion-native ways. I do not have definitive answers, but I do point my fingers in directions I deem interesting. A tutorial follows this thesis, expanding on the destroy-then-generate perspective. A novel kind of probabilistic graphical models is introduced to facilitate the tutorial's exposition. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_30553 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Destruction is a General Strategy to Learn Generation; Diffusion's Strength is to Take it Seriously; Exploration is the Future Noël, Pierre-André Machine Learning Information Theory I present diffusion models as part of a family of machine learning techniques that withhold information from a model's input and train it to guess the withheld information. I argue that diffusion's destroying approach to withholding is more flexible than typical hand-crafted information withholding techniques, providing a rich training playground that could be advantageous in some settings, notably data-scarce ones. I then address subtle issues that may arise when porting reinforcement learning techniques to the diffusion context, and wonder how such exploration problems could be addressed in more diffusion-native ways. I do not have definitive answers, but I do point my fingers in directions I deem interesting. A tutorial follows this thesis, expanding on the destroy-then-generate perspective. A novel kind of probabilistic graphical models is introduced to facilitate the tutorial's exposition. |
| title | Destruction is a General Strategy to Learn Generation; Diffusion's Strength is to Take it Seriously; Exploration is the Future |
| topic | Machine Learning Information Theory |
| url | https://arxiv.org/abs/2605.30553 |