PropInit: Scalable Inductive Initialization for Heterogeneous Graph Neural Networks.
Soji AdeshinaJian ZhangMuhyun KimMin ChenRizal FathonyAdvitiya VashishtJia ChenGeorge KarypisPublished in: ICKG (2022)
Keyphrases
- neural network
- pattern recognition
- graph representation
- graph theory
- artificial neural networks
- weighted graph
- random walk
- graph structure
- graph model
- back propagation
- graph clustering
- directed graph
- fuzzy systems
- concept learning
- fuzzy logic
- genetic algorithm
- machine learning
- neural network model
- highly scalable
- bipartite graph
- self organizing maps
- structured data
- knowledge representation
- training process
- web scale
- inductive logic programming
- graph based algorithm
- graph theoretic
- multilayer perceptron
- radial basis function
- fault diagnosis
- real world