Classical neural architectures as GNN special cases

Early works established that neural network models such as perceptrons, feedforward nets, CNNs, and RNNs can be interpreted as special instances of generalized GNNs applied to specific graph structures such as grids or chains.

Part of Chain