Learning the hypotheses space from data through a U-curve algorithm: a statistically consistent complexity regularizer for Model Selection.
Diego MarcondesAdilson SimonisJunior BarreraPublished in: CoRR (2021)
Keyphrases
- model selection
- bayesian methods
- learning algorithm
- data sets
- automatic model selection
- worst case
- parameter estimation
- hyperparameters
- k means
- cross validation
- statistical learning
- prior knowledge
- expectation maximization
- bayesian learning
- supervised learning
- objective function
- incremental learning
- training data
- gaussian process
- incomplete data
- motion segmentation
- statistical inference
- marginal likelihood
- data points
- multi task learning
- regression model
- bayesian framework
- energy function
- high dimensional data
- em algorithm
- machine learning