Calibrate Graph Neural Networks under Out-of-Distribution Nodes via Deep Q-learning.
Weili ShiXueying YangXujiang ZhaoHaifeng ChenZhiqiang TaoSheng LiPublished in: CIKM (2023)
Keyphrases
- neural network
- graph structure
- directed graph
- weighted graph
- undirected graph
- pattern recognition
- edge weights
- adjacency matrix
- spanning tree
- multi agent
- reinforcement learning
- cooperative
- function approximation
- small world
- graph structures
- directed edges
- dominating set
- root node
- random walk
- connected graphs
- nodes of a graph
- power law
- fully connected
- input pattern
- planar graphs
- real world graphs
- average degree
- artificial neural networks
- probability distribution
- fuzzy logic
- minimum cost
- genetic algorithm
- finding the shortest path
- state space
- strongly connected
- camera calibration
- random graphs
- graph matching
- attributed graphs
- bipartite graph
- betweenness centrality
- graph theory
- graph model
- learning algorithm
- stochastic approximation
- neural network model
- sufficient conditions
- degree distribution
- massive graphs
- directed acyclic graph
- graph partitioning