A Meta-Reinforcement Learning Approach to Optimize Parameters and Hyper-parameters Simultaneously.
Abbas Raza AliMarcin BudkaBogdan GabrysPublished in: PRICAI (2) (2019)
Keyphrases
- hyperparameters
- model selection
- reinforcement learning
- cross validation
- closed form
- bayesian inference
- bayesian framework
- maximum likelihood
- support vector
- random sampling
- prior information
- em algorithm
- sample size
- variational bayes
- maximum a posteriori
- gaussian process
- incremental learning
- bayesian methods
- noise level
- parameter optimization
- posterior distribution
- parameter space
- parameter settings
- parameter estimation
- expectation maximization
- state space
- incomplete data
- learning algorithm
- image reconstruction
- lower bound
- high resolution
- machine learning