Neural Tangent Kernels Motivate Graph Neural Networks with Cross-Covariance Graphs.
Shervin KhalafiSaurabh SihagAlejandro RibeiroPublished in: CoRR (2023)
Keyphrases
- neural network
- graph kernels
- labeled graphs
- graph representation
- graph structure
- network architecture
- graph classification
- graph theory
- graph matching
- weighted graph
- graph mining
- graph construction
- graph databases
- graph theoretic
- directed graph
- adjacency matrix
- graph structures
- graph partitioning
- graph representations
- graph properties
- graph model
- structural pattern recognition
- graph clustering
- graph search
- graph data
- pattern recognition
- series parallel
- random graphs
- kernel function
- graph isomorphism
- connected graphs
- undirected graph
- artificial neural networks
- spanning tree
- reachability queries
- graph theoretical
- subgraph isomorphism
- bipartite graph
- dynamic graph
- minimum spanning tree
- neural model
- graph patterns
- associative memory
- graph embedding
- connectionist models
- neural fuzzy
- neural learning
- query graph
- kernel methods
- structured data
- massive graphs
- real world graphs
- edge weights
- learning rules
- attributed graphs
- pattern mining
- graph layout
- social networks
- maximal cliques
- community discovery
- random walk
- web graph
- feed forward
- maximum common subgraph
- finding the shortest path
- covariance matrix
- neural network model
- polynomial time complexity
- small world
- adjacency graph
- graph laplacian
- recurrent neural networks
- neighborhood graph
- pairwise
- heat kernel