Confidence May Cheat: Self-Training on Graph Neural Networks under Distribution Shift.
Hongrui LiuBinbin HuXiao WangChuan ShiZhiqiang ZhangJun ZhouPublished in: WWW (2022)
Keyphrases
- neural network
- pattern recognition
- semi supervised classification
- graph theory
- genetic algorithm
- probability distribution
- fuzzy logic
- random walk
- co training
- graph structure
- structured data
- connected components
- multilayer perceptron
- high confidence
- spatial distribution
- recurrent neural networks
- semi supervised learning
- data distribution
- back propagation
- training set
- multi layer
- graph matching
- power law
- graph representation
- graph model
- degree distribution
- training data
- confidence measure
- small world
- support vector machine
- graph mining
- weighted graph
- neural nets
- neural network model
- data sets