AGNN: Alternating Graph-Regularized Neural Networks to Alleviate Over-Smoothing.
Zhaoliang ChenZhihao WuZhenghong LinShiping WangClaudia PlantWenzhong GuoPublished in: CoRR (2023)
Keyphrases
- neural network
- fuzzy logic
- graph representation
- graph theory
- pattern recognition
- random walk
- neural network model
- solution path
- artificial neural networks
- structured data
- connected components
- graph matching
- least squares
- graph structure
- back propagation
- self organizing maps
- directed graph
- feed forward
- image processing
- graph model
- directed acyclic graph
- regularization method
- graph theoretic
- graph construction
- multiscale
- graph based algorithm
- genetic algorithm
- spanning tree
- graph databases
- weighted graph
- bipartite graph
- objective function
- fault diagnosis
- edge detection