Alignment and Safety of Diffusion Models via Reinforcement Learning and Reward Modeling: A Survey
Preeti Lamba, Kiran Ravish, Ankita Kushwaha, Pawan Kumar · 2025
Read Alignment and Safety of Diffusion Models via Reinforcement Learning and Reward Modeling: A Survey by Preeti Lamba, Kiran Ravish, Ankita Kushwaha, Pawan Kumar (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-17352.txt.
Description
Preeti Lamba and co-authors survey how researchers steer image-making diffusion models toward what people want and away from harm. Their 2025 review covers methods that train these models with rewards and human feedback. The field has grown fast, and the paper aims to give it a clear map.
The authors sort the work by where feedback comes from, what form the reward takes, and how training uses it. They compare human feedback training, preference tuning, and safety-focused variants on cost, scale, and risk of gaming the reward. Tutorial sections explain sampling and reward models in plain terms. Open problems include balancing many goals at once and keeping safety strong as norms change.
Catalog record
| CREATOR | Preeti Lamba, Kiran Ravish, Ankita Kushwaha, Pawan Kumar |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 83 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-17352