In this paper, Graph-MLP Mixer, the authors demonstrate a successful generalization of the ViT/MLP-Mixer architectures introduced in computer vision for graph representation learning:
Xavier Bresson@xbresson
New paper w/ @he_xiaoxin @BryanHooi1 T Laurent @AdamPerold @ylecun arxiv.org/pdf/2212.13350… We introduce Graph MLP-Mixer, a GNN w/ three key properties: 1) captures long-range dependency 2) keeps low linear speed/memory complexity as MP-GNNs 3) provides high expressivity
English
