- Article
Despite the success of Geometric Graph Neural Networks (GGNNs), their reliance on local message passing inherently limits the receptive field, leading to the over-squashing of distant information. Current solutions—ranging from graph rewiring t...