MolGrow: A Graph Normalizing Flow for Hierarchical Molecular Generation

We propose a hierarchical normalizing flow model for generating molecular graphs. The produces new structures from single-node graph by recursively splitting every node into two. All operations are invertible and can be used as plug-and-play modules. nature of the latent codes allows precise changes in resulting graph: perturbations first layer cause global structural changes, while consequent layers change molecule only marginally. Proposed outperforms existing generative models on distribution learning task. also show successful experiments constrained optimization chemical properties using model.