Active Heterogeneous Graph Neural Networks with Per-step Meta-Q-Learning.
Yuheng ZhangYinglong XiaYan ZhuYuejie ChiLei YingHanghang TongPublished in: ICDM (2022)
Keyphrases
- neural network
- reinforcement learning
- graph theory
- function approximation
- cooperative
- pattern recognition
- graph transformation
- multilayer perceptron
- fuzzy logic
- fuzzy systems
- graph representation
- preprocessing stage
- post processing
- directed graph
- learning rate
- bipartite graph
- multi agent reinforcement learning
- machine learning
- artificial neural networks
- learning algorithm
- feed forward
- neural nets
- connected components
- structured data
- back propagation
- graph structure
- weighted graph
- random walk
- state space
- directed acyclic graph
- graph model
- meta level
- multi agent
- social networks