Deep learning on graphs has become a popular research topic with many. However, past work has concentrated on learning graph embedding, which is in contrast with advances in generative models for images and. Is it possible to transfer this progress to the domain of graphs? We to sidestep hurdles associated with linearization of such discrete by having a decoder output a probabilistic fully-connected graph of predefined maximum size directly at once. Our method is formulated as a autoencoder. We evaluate on the challenging task of molecule.
No takes yet. Share an insight, caveat, or question.
Simonovsky et al. (2018) studied this question.