MPGVAE: improved generation of small organic molecules using message passing neural nets
Daniel Flam-Shepherd, Tony Wu, Alán Aspuru‐Guzik · Machine Learning Science and Technology · 2021
Abstract Graph generation is an extremely important task, as graphs are found throughout different areas of science and engineering. In this work, we focus on the modern equivalent of the Erdos–Rényi random graph model: the graph variational autoencoder (GVAE) (Simonovsky and Komodakis 2018 Int. Conf. on Artificial Neural Networks pp 412–22). This model assumes edges and nodes are independent in order to generate entire graphs at a time using a multi-layer perceptron decoder. As a result of these assumptions, GVAE has difficulty matching the training distribution and relies on an expensive graph matching procedure. We improve this class of models by building a message passing neural network into GVAE’s encoder and decoder. We demonstrate our model on the specific task of generating small organic molecules.