Bridging Smoothness and Approximation: Theoretical Insights into Over-Smoothing in Graph Neural Networks.
Guangrui YangJianfei LiMing LiHan FengDing-Xuan ZhouPublished in: CoRR (2024)
Keyphrases
- theoretical insights
- neural network
- smoothing parameter
- pattern recognition
- graph representation
- graph model
- bipartite graph
- graph matching
- directed graph
- graph mining
- multi layer
- weighted graph
- feed forward
- graph structure
- cost function
- back propagation
- graph theoretic
- artificial neural networks
- neural nets
- minimum spanning tree
- recurrent neural networks
- knn
- error bounds
- random walk
- adjacency matrix
- structured data
- approximation error
- gaussian convolution
- graph partitioning
- approximation algorithms
- neural network model
- connected components
- self organizing maps
- fuzzy logic
- genetic algorithm