复杂认知图神经网络综述
1. 面向复杂图的图神经网络
- 面向通用网络的图神经网络:考虑不同网络属性(如社交网络和拍卖网络)的不同传播机制,提出了一种通用图卷积网络(Universal Graph Convolutional Networks, UGCN),以适应不同类型的网络。
- 面向文本丰富的网络的图神经网络:提出了一种双类型网络构建方法(BiTe-GCN),通过结合拓扑结构和特征信息,改进了文本丰富的网络处理。
- 面向属性缺失的异质信息网络的图神经网络:提出了通过属性完成的方法来处理节点属性缺失的问题。
- 面向高阶依赖关系网络的图神经网络:结合马尔可夫随机场(Markov Random Fields, MRF)和图卷积网络(Graph Convolutional Networks, GCN),提出了一个端到端的学习框架。
2. 认知图神经网络
- 随机游走聚合的图神经网络(RAW-GNN):通过随机游走聚合机制,提高了图神经网络在处理高阶依赖关系网络中的表现。
- 对比图匹配网络(CGMN):用于自监督的图相似性学习。
- 高阶依赖关系网络的图神经网络:结合马尔可夫随机场和图卷积网络,提出了一种新的学习框架。
关键研究成果
- 2022年主要成果:
- RAW-GNN: RAndomWalk Aggregation based Graph Neural Network
- CGMN: A Contrastive Graph Matching Network for Self-Supervised Graph Similarity Learning
- Graph Neural Network for Higher-Order Dependency Networks
- A New Attribute-Missing Network Embedding Approach
- Powerful Graph Convolutioal Networks with Adaptive Propagation Mechanism
- Fast Algorithms for Core Maximization on Large Graphs
- Analyzing Heterogeneous Networks with Missing Attributes by Unsupervised Contrastive Learning
- GATrust: A Multi-Aspect Graph Attention Network Model for Trust Assessment in OSNs
- 2021年主要成果:
- Heterogeneous Graph Neural Network via Attribute Completion
- AS-GCN: Adaptive Semantic Architecture of Graph Convolutional Networks for Text-Rich Networks
- GCN for HIN via Implicit Utilization of Attention and Meta-paths
- Robust Detection of Link Communities with Summary Description in Social Networks
- Universal Graph Convolutional Networks
- BiTe-GCN: A New GCN Architecture via Bidirectional Convolution of Topology and Features on Text-rich Networks
- Budget-constrained Truss Maximization over Large Graphs: A Component-based Approach
- Self-Guided Community Detection on Networks with Missing Edges
- Adversarial Representation Mechanism Learning for Network Embedding
以上总结涵盖了复杂图神经网络的关键内容和重要研究成果。