We study the problem of learning conditional distributions of the form p(G∣G^), where G and G^ are two 3D graphs, using continuous normalizing flows. We derive a semi-equivariance condition on the flow which ensures that conditional invariance to rigid motions holds. We demonstrate the effectiveness of the technique in the molecular setting of receptor-aware ligand generation.
Cite
@article{arxiv.2304.06779,
title = {Semi-Equivariant Conditional Normalizing Flows},
author = {Eyal Rozenberg and Daniel Freedman},
journal= {arXiv preprint arXiv:2304.06779},
year = {2023}
}
Comments
ICLR Physics for Machine Learning (Physics4ML) Workshop 2023. arXiv admin note: substantial text overlap with arXiv:2211.04754