Data-Driven Regret Balancing for Online Model Selection in Bandits.
Aldo PacchianoChristoph DannClaudio GentilePublished in: CoRR (2023)
Keyphrases
- model selection
- data driven
- online learning
- cross validation
- hyperparameters
- parameter estimation
- regret bounds
- machine learning
- sample size
- regression model
- statistical learning
- meta learning
- model selection criteria
- online convex optimization
- generalization error
- mixture model
- bayesian learning
- error estimation
- selection criterion
- feature selection
- motion segmentation
- lower bound
- variable selection
- statistical inference
- information criterion
- gaussian process
- loss function
- bayesian information criterion
- automatic model selection
- multi armed bandit problems
- feature selection algorithms
- generalization bounds
- worst case
- leave one out cross validation
- support vector
- subspace information criterion