Information-theoretic generalization bounds for black-box learning algorithms.
Hrayr HarutyunyanMaxim RaginskyGreg Ver SteegAram GalstyanPublished in: NeurIPS (2021)
Keyphrases
- information theoretic
- black box
- generalization bounds
- generalization ability
- learning algorithm
- learning problems
- algorithmic stability
- learning theory
- data dependent
- information theory
- mutual information
- kernel machines
- machine learning algorithms
- learning machines
- statistical learning theory
- vc dimension
- model selection
- ranking algorithm
- machine learning
- learning tasks
- test cases
- supervised learning
- sample complexity
- reinforcement learning
- support vector
- active learning
- linear classifiers
- support vector machine svm
- semi supervised learning
- generalization error
- ranking functions
- training data
- support vector machine
- decision trees
- hyperparameters
- kernel methods
- pattern recognition
- transfer learning
- image analysis