HyperDrive: exploring hyperparameters with POP scheduling.
Jeff RasleyYuxiong HeFeng YanOlatunji RuwaseRodrigo FonsecaPublished in: Middleware (2017)
Keyphrases
- hyperparameters
- model selection
- cross validation
- closed form
- bayesian framework
- bayesian inference
- support vector
- em algorithm
- noise level
- gaussian process
- maximum likelihood
- random sampling
- maximum a posteriori
- prior information
- incremental learning
- gaussian processes
- incomplete data
- sample size
- regularization parameter
- grid search
- machine learning
- missing values
- markov random field
- upper bound
- active learning
- parameter settings
- expectation maximization
- learning algorithm