Parallelizing Graph Neural Networks via Matrix Compaction for Edge-Conditioned Networks.
Shehtab ZamanTim MoonTom BensonSam Adé JacobsKenneth ChiuBrian Van EssenPublished in: CCGRID (2022)
Keyphrases
- neural network
- edge weights
- weighted graph
- adjacency matrix
- undirected graph
- bipartite graph
- pattern recognition
- directed graph
- hopfield neural network
- edge detection
- fuzzy logic
- disjoint paths
- community discovery
- dynamic networks
- graph partitioning
- social networks
- graph theory
- edge information
- eigenvalues and eigenvectors
- small world
- fully connected
- genetic algorithm
- vertex set
- artificial neural networks
- complex networks
- transition matrix
- random walk
- linear algebra
- graph clustering
- graph model
- structured data
- overlapping communities
- feed forward
- graph structures
- neural network ensemble
- similarity matrix
- graph matching
- graph representation
- network analysis
- densely connected
- betweenness centrality
- directed edges
- laplacian matrix
- back propagation
- clustering coefficient
- network structure
- parallel processing
- power law
- graph structure
- community detection
- graph databases
- directed acyclic graph