Path Integral Based Convolution and Pooling for Graph Neural Networks.
Zheng MaJunyu XuanYu Guang WangMing LiPietro LiòPublished in: NeurIPS (2020)
Keyphrases
- neural network
- pattern recognition
- solution path
- shortest path
- graph structure
- graph representation
- graph theory
- random walk
- fuzzy logic
- genetic algorithm
- integral equation
- shortest path problem
- weighted graph
- feed forward
- artificial neural networks
- image processing
- path length
- back propagation
- directed graph
- multilayer perceptron
- graph matching
- strongly connected
- graph theoretic
- edge weights
- radial basis function
- source node
- multiscale
- graph construction
- labeling scheme
- pairwise
- bipartite graph
- neural network model
- activation function
- graph partitioning
- directed acyclic graph
- connected components