Shift-Robust GNNs: Overcoming the Limitations of Localized Graph Training data.
Qi ZhuNatalia PonomarevaJiawei HanBryan PerozziPublished in: NeurIPS (2021)
Keyphrases
- training data
- data sets
- classification accuracy
- random walk
- graph structure
- test data
- supervised learning
- graph theory
- graph representation
- support vector machine
- partial occlusion
- graph mining
- graph model
- graph databases
- decision trees
- machine learning
- robust estimation
- label noise
- graph theoretic
- training instances
- graph structures
- connected components
- structured data
- test set
- semi supervised
- prior knowledge
- multiresolution
- training set
- similarity measure
- learning algorithm