Combined l1 and greedy l0 penalized least squares for linear model selection.
Piotr PokarowskiJan MielniczukPublished in: J. Mach. Learn. Res. (2015)
Keyphrases
- model selection
- least squares
- parameter estimation
- cross validation
- feature selection
- regression model
- hyperparameters
- bayesian learning
- variable selection
- statistical learning
- sample size
- meta learning
- linear regression
- generalization error
- statistical inference
- information criterion
- machine learning
- model selection criteria
- optical flow
- leave one out cross validation
- marginal likelihood
- gaussian process
- bayesian methods
- error estimation
- bayesian information criterion
- automatic model selection
- similarity measure