Fast and Deep Graph Neural Networks (FDGNN) is a novel model for learning in graph domains featured by an extremely fast training algorithm.Â
The research paper has been accepted for presentation at the 34th AAAI Conference on Artificial Intelligence, 2020.
A pre-print version, including the supporting material (with proofs of the theorems in the paper) can be found here.
The code is made publicly available on this dedicated page.