Evolutionary training-free guidance in diffusion model for 3D multi-objective molecular generation
Ruiqing Sun, Dawei Feng, Sen Yang, Yijie Wang, Huaimin Wang · 2025
Read Evolutionary training-free guidance in diffusion model for 3D multi-objective molecular generation by Ruiqing Sun, Dawei Feng, Sen Yang, Yijie Wang, Huaimin Wang (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-2505-11037.txt.
Description
Molecules can be designed by evolution inside a frozen diffusion model, in a 2025 paper by Ruiqing Sun and co-authors. The task is to invent 3D ones meeting several property goals while keeping chosen fragments intact, a constrained multi-objective puzzle.
Their framework, DEMO, uses only the prior of the frozen model, never retraining it. The evolution-guided operator crosses and mutates the model's midway states, while a structure-aware selection step keeps shapes and topology diverse. Three cooperating populations divide the labor: exploration, fragment keeping, and goal chasing. Tests report the method beating task-specific conditional generators. The design keeps the learned chemistry apart from the search, so new goals need no new training.
Catalog record
| CREATOR | Ruiqing Sun, Dawei Feng, Sen Yang, Yijie Wang, Huaimin Wang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-11037