combinations = [('GCN', 'GAT'), ('GCN', 'GIN'), ('GAT', 'GIN')] fig, axes = plt.subplots(1, 3, figsize=(15, 4),dpi=128) for ax, (model1, model2) in zip(axes, combinations): scatter_data = pd.DataFrame({ f'{model1}_Probabilities': temp[temp['Model'] == model1][...
Spatial-based GNN部分,首先复习CNN的filter计算;接着介绍了各种技术:NN4G(Neural Networks for Graph),DCNN(Diffusion-Convolution Neural Network),MoNET(Mixture Model Networks),GraphSAGE(SAmple and aggreGatE),GAT(Graph Attention Networks),GIN(Graph Isomorphism Network)。 Graph Signal Processing and Spectral-ba...