An Empirical Investigation of Pre-trained Model Selection for Out-of-Distribution Generalization and Calibration.
Hiroki NaganumaRyuichiro HatayaPublished in: CoRR (2023)
Keyphrases
- model selection
- pre trained
- cross validation
- generalization bounds
- model selection criteria
- machine learning
- mixture model
- training examples
- training data
- generalization error
- feature selection
- gaussian process
- selection criterion
- unsupervised learning
- probability distribution
- information criterion
- learning algorithm
- state space
- marginal likelihood
- neural network