A PAC-Bayesian Approach to Generalization Bounds for Graph Neural Networks.
Renjie LiaoRaquel UrtasunRichard S. ZemelPublished in: ICLR (2021)
Keyphrases
- generalization bounds
- vc dimension
- neural network
- pac bayesian
- data dependent
- learning theory
- generalization ability
- model selection
- sample complexity
- upper bound
- learning problems
- statistical learning theory
- ranking algorithm
- linear classifiers
- lower bound
- sample size
- pattern recognition
- fuzzy logic
- neural network model
- special case
- uniform convergence
- concept classes
- inductive inference
- multi class
- artificial neural networks
- learning machines
- worst case
- kernel machines
- ranking functions
- back propagation
- learning tasks
- feature space
- learning algorithm