The generalization error of graph convolutional networks may enlarge with more layers.
Xianchen ZhouHongxia WangPublished in: Neurocomputing (2021)
Keyphrases
- generalization error
- cross validation
- model selection
- upper bound
- active learning
- sample complexity
- training set
- binary classification
- training data
- training error
- classification error
- learning algorithm
- supervised learning
- linear classifiers
- target function
- sample size
- generalization error bounds
- subspace information criterion
- learning machines
- perceptron learning
- semi supervised
- reinforcement learning
- restricted boltzmann machine
- deep belief networks