GCNII

GCNII is an extension of a Graph Convolution Networks with two new techniques, initial residual and identify mapping, to tackle the problem of oversmoothing -- where stacking more layers and adding non-linearity tends to degrade performance. At each layer, initial residual constructs a skip connection from the input layer, while identity mapping adds an identity matrix to the weight matrix.
相关学科: AdaGPRS-GCNGATGCNGraph ClassificationGraph AttentionNode ClassificationKnowledge DistillationLink PredictionConvolution

学科讨论

讨论Icon

暂无讨论内容,你可以

推荐文献

按被引用数

学科管理组

暂无学科课代表,你可以申请成为课代表

重要学者

Chuan Shi

4731 被引用,212 篇论文

Yaliang Li

2704 被引用,142 篇论文

Cheng Yang

2585 被引用,60 篇论文

Zhewei Wei

1037 被引用,68 篇论文

Zengfeng Huang

665 被引用,68 篇论文

Bolin Ding

509 被引用,131 篇论文

Ming Chen

196 被引用,6 篇论文

Jiawei Liu

10 被引用,2 篇论文