Diffusion-Driven Domain Adaptation for Generating 3D Molecules

arXiv:2404.00962v1 Announce Type: new
Abstract: Can we train a molecule generator that can generate 3D molecules from a new domain, circumventing the need to collect data? This problem can be cast as the problem of domain adaptive molecule generation. This work presents a novel and principled diffusion-based approach, called GADM, that allows shifting a generative model to desired new domains without the need to collect even a single molecule. As the domain shift is typically caused by the structure variations …

abstract arxiv cs.lg data diffusion domain domain adaptation generate generator molecules novel physics.chem-ph q-bio.bm train type work