Graph Energy-based Model for Molecular Graph Generation

Ryuichiro Hataya, Hideki Nakayama, Kazuki Yoshizoe

Keywords: molecular graph generation

TL;DR: We propose GEM, an EBM-based molecular graph generation method

Abstract: We present Graph Energy-based Model (GEM), an energy-based model for molecular graph generation. GEM uses dequantization and gradient symmetrization to incorporate generation by stochastic gradient Langevin dynamics for graph representation that is discrete and includes symmetric constraint. Experimental results show that \gem can comparably design compounds as other deep generative approaches.