Molecule Property Prediction and Classification with Graph Hypernetworks

--Graph neural networks are currently leading the performance charts in learning-based molecule property prediction and classification. Computational chemistry has, therefore, become the a prominent testbed for generic graph neural networks, as well as for specialized message passing methods. In this work, we demonstrate that the replacement of the underlying networks with hypernetworks leads to a boost in performance, obtaining state of the art results in various benchmarks. A major difficulty in the application of hypernetworks is their lack of stability. We tackle this by combining the current message and the first message. A recent work has tackled the training instability of hypernetworks in the context of error correcting codes, by replacing the activation function of the message passing network with a low-order T aylor approximation of it. We demonstrate that our generic solution can replace this domain-specific solution. I NTRODUCTION The field of learning-based prediction of molecule properties holds the promise of delivering accurate predictions at a fraction of the complexity that is required by the Density Functional Theory (DFT) models, while not being tied to the assumptions and approximations of this theory.