Generic Graph Convolutional Neural Network (GCN) for Molecular and Solid-state Materials

This model is built on top of the original CGCNN implementation:

[email protected]:txie-93/cgcnn.git

The aim of this research project is to develop generic materials representations and the corresponding neural network architectures (e.g. multigraph) capable of learning the structure-property relationship in molecular and condensed-matter materials systems.