A multi-objective perspective on jointly tuning hardware and hyperparameters.
David SalinasValerio PerroneOlivier CruchantCédric ArchambeauPublished in: CoRR (2021)
Keyphrases
- hyperparameters
- multi objective
- model selection
- cross validation
- closed form
- bayesian inference
- parameter settings
- support vector
- bayesian framework
- gaussian process
- prior information
- evolutionary algorithm
- em algorithm
- random sampling
- sample size
- maximum a posteriori
- gaussian processes
- maximum likelihood
- noise level
- incremental learning
- regularization parameter
- objective function
- incomplete data
- genetic algorithm
- missing values
- parameter optimization
- probabilistic model
- machine learning
- parameter space
- markov random field
- prior knowledge
- signal to noise ratio
- worst case
- upper bound
- training set
- feature selection
- data mining
- grid search