图神经网络汇总
图神经基础
torch-geometric
torch_geometric API
DGL
DGL API
图神经Embedding
-
2014 - DeepWalk: Online Learning of Social Representations
DeepWalk:采用随机游走,形成序列,采用skip-gram方式生成节点embedding。
原理
代码 -
2015 - LINE: Large-scale Information Network Embedding
LINE:捕获节点的一阶和二阶相似度,分别求解,再将一阶二阶拼接在一起,作为节点的embedding
原理
代码 -
2016 - node2vec:Scalable Feature Learning for Networks
node2vec:不同的随机游走策略,形成序列,类似skip-gram方式生成节点embedding。
原理
代码 -
2016 - SDNE: Structural Deep Network Embedding
SDNE:采用了多个非线性层的方式捕获一阶二阶的相似性。
原理
代码 -
2017 - struc2vec: Learning Node Representations from Structural Identity
struc2vec:对图的结构信息进行捕获,在其结构重要性大于邻居重要性时,有较好的效果。
原理
代码 -
2017 - metapath2vec: Scalable Representation Learning for Heterogeneous
metapath2vec:基于metapath的随机游走策略, gensim+word2vec来实现embedding。
原理
代码
图卷积
-
2017 - GCN: Semi-Supervised Classification with Graph Convolutional Networks
原理
代码 -
2017 - GraphSAGE : Inductive Representation Learning on Large Graphs
原理
代码 -
2017 - GAT: Semi-Supervised Classification with Graph Convolutional Networks
原理
代码 -
2018 - RGCN: Modeling Relational Data with Graph Convolutional Networks
原理
代码 -
2019 - GTN:Graph Transformer Networks
原理
代码
2017-PMLR-Neural Message Passing for Quantum Chemistry
2019-LCLR-How Powerful are Graph Neural Networks
图池化
- 2018-NIPS-owards Sparse Hierarchical Graph Classifiers
- 2018-NeurIPS-Hierarchical Graph Representation Learning with Differentiable Pooling
- 2019-ICML-Graph U-Nets
- 2019-NeurIPS-Understanding Attention and Generalization in Graph Neural Networks
- 2019-ICML-Towards Graph Pooling by Edge Contraction
- 2019-arXiv-Edge Contraction Pooling for Graph Neural
- 2020-ICLR-Memory-Based Graph Networks
- 2020-AAAI-ASAP : Adaptive Structure Aware Pooling for Learning Hierarchical Graph Representations
- 2020-NeurIPS-Path Integral Based Convolution and Pooling for Graph Neural Networks
- 2020-ICML-Spectral Clustering with Graph Neural Networks for Graph Pooling
图Readout
图生成
图可解释性
- GNNExplainer
- PGExplainer
- ProtGNN