过平滑问题和深层GNN优质论文合集

Oversmoothing and Deep GNNs：研究GNN的过平滑问题和深层GNN。GNN的卷积操作在层数过多时会使得每个结点几乎都能间接融合其它所有结点的信息，导致所有结点的表征趋于相同，因此通常GNN的层数不宜设置过大。例如：推荐场景中的GNN通常设置为1层或2层即可达到最佳性能。如何突破这种瓶颈，使得GNN的层数更深，又不会导致过度平滑问题，性能更好，也是近年来比较热门的研究方向。
小编整理了近期过平滑问题和深层GNN主题论文，现分享给大家~

1.论文名称：Representation Learning on Graphs with Jumping Knowledge Networks.
链接：https://www.aminer.cn/pub/5b67b47917c44aac1c8637c6
2.论文名称：Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning.
链接：https://www.aminer.cn/pub/5a9cb66717c44a376ffb8c0f
3.论文名称：Predict then Propagate: Graph Neural Networks meet Personalized PageRank
链接：https://www.aminer.cn/pub/5ce2d032ced107d4c635260c
4.论文名称：DeepGCNs: Can GCNs Go As Deep As CNNs?
链接：https://www.aminer.cn/pub/5e09aa66df1a9c0c416bebf6
5.论文名称：Layer-Dependent Importance Sampling for Training Deep and Large Graph Convolutional Networks
链接：https://www.aminer.cn/pub/5db9292e47c8f766461eff27
6.论文名称：DeeperGCN: All You Need to Train Deeper GCNs
链接：https://www.aminer.cn/pub/5ee8986891e011e66831c3bc
7.论文名称：PairNorm: Tackling Oversmoothing in GNNs
链接：https://www.aminer.cn/pub/5e5e18d593d709897ce3398b
8.论文名称：DropEdge: Towards Deep Graph Convolutional Networks on Node Classification
链接：https://www.aminer.cn/pub/5e5e18de93d709897ce37796
9.论文名称：Simple and Deep Graph Convolutional Networks
链接：https://www.aminer.cn/pub/5ede0553e06a4c1b26a83f63
10.论文名称：Towards Deeper Graph Neural Networks
链接：https://www.aminer.cn/pub/5f03f3b611dc83056223205d
11.论文名称：Towards Deeper Graph Neural Networks with Differentiable Group Normalization
链接：https://www.aminer.cn/pub/5ee7495191e01198a507f880
12.论文名称：Scattering GCN: Overcoming Oversmoothness in Graph Convolutional Networks
链接：https://www.aminer.cn/pub/5e7495c591e0111c7cee12e7
13.论文名称：Optimization and Generalization Analysis of Transduction through Gradient Boosting and Application to Multi-scale Graph Neural Networks
链接： https://www.aminer.cn/pub/5ee8986f91e011e66831c6e7
14.论文名称：On the Bottleneck of Graph Neural Networks and its Practical Implications
链接：https://www.aminer.cn/pub/5ee3526a91e011cb3bff715d
15.论文名称：Simple Spectral Graph Convolution
链接：https://www.aminer.cn/pub/600832489e795ed227f530f8
16.论文名称：Training Graph Neural Networks with 1000 Layers
链接：https://www.aminer.cn/pub/60cae12b91e011b32937419e
17.论文名称：Optimization of Graph Neural Networks: Implicit Acceleration by Skip Connections and More Depth
链接：https://www.aminer.cn/pub/609a3ab091e011a44725cc5b
18.论文名称：GRAND: Graph Neural Diffusion
链接：https://www.aminer.cn/pub/60d3fe7891e0112ca5d18779
19.论文名称：Directional Graph Networks
链接：https://www.aminer.cn/pub/5f7d9dea91e011346ad27f48
20.论文名称：Improving Breadth-Wise Backpropagation in Graph Neural Networks helps Learning Long-Range Dependencies.
链接：https://www.aminer.cn/pub/60bdde338585e32c38af51dd
更多优质论文，尽在AMiner，主页添加关键词，系统智能推荐最新优质论文~
AMiner平台链接：https://www.aminer.cn/?f=cs