ReGNN: A Redundancy-Eliminated Graph Neural Networks Accelerator.
Cen ChenKenli LiYangfan LiXiaofeng ZouPublished in: HPCA (2022)
Keyphrases
- neural network
- graph representation
- random walk
- graph theory
- graph structure
- graph model
- structured data
- pattern recognition
- back propagation
- fuzzy logic
- graph partitioning
- artificial neural networks
- graph construction
- connected components
- neural network model
- multilayer perceptron
- genetic algorithm
- weighted graph
- stable set
- graph clustering
- rule extraction
- graph theoretic
- graph mining
- graph matching
- recurrent neural networks
- information content
- fault diagnosis
- dependency graph
- graph structures
- graph search
- self organizing maps
- directed graph
- training process
- expert systems
- parallel implementation
- graph databases
- network architecture
- data sets
- fuzzy systems
- directed acyclic graph