Semi-Equivariant Conditional Normalizing Flows

Eyal Rozenberg, Daniel Freedman

Keywords: 3D graphs conditional distributions, conditional invariance, continuous normalizing flows, semi-equivariance

TL;DR: We derive semi-equivariance conditions to yield invariant conditional distributions between two 3D graphs, using continuous normalizing flows.

Abstract: We study the problem of learning conditional distributions of the form $p(G | \hat{G})$, where $G$ and $\hat{G}$ 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.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/semi-equivariant-conditional-normalizing/code)