Refining Latent Homophilic Structures over Heterophilic Graphs for Robust Graph Convolution Networks.
Chenyang QiuGuoshun NanTianyu XiongWendi DengDi WangZhiyang TengLijuan SunQimei CuiXiaofeng TaoPublished in: AAAI (2024)
Keyphrases
- graph structures
- graph structure
- average degree
- graph representation
- directed graph
- edge weights
- weighted graph
- highly connected
- community discovery
- graph clustering
- graph mining
- graph databases
- graph layout
- dynamic networks
- graph theory
- graph search
- fully connected
- graph theoretic
- graph matching
- graph model
- structured data
- undirected graph
- graph data
- adjacency matrix
- real world graphs
- bipartite graph
- small world
- random graphs
- social networks
- graph construction
- densely connected
- real world social networks
- random walk
- graphical models
- introduce a general framework
- complex networks
- graph classification
- labeled graphs
- graph theoretical
- real world networks
- reachability queries
- minimum spanning tree
- complex structures
- spanning tree
- directed edges
- series parallel
- shortest path
- degree distribution
- structural pattern recognition
- network analysis
- biological networks
- dense subgraphs
- community detection
- web graph
- massive graphs
- connected components
- subgraph isomorphism
- adjacency graph
- graph properties
- relational structures
- structural patterns
- attributed graphs
- protein interaction networks