Optimizing Millions of Hyperparameters by Implicit Differentiation.
Jonathan LorrainePaul VicolDavid DuvenaudPublished in: AISTATS (2020)
Keyphrases
- hyperparameters
- model selection
- cross validation
- closed form
- bayesian inference
- random sampling
- bayesian framework
- support vector
- em algorithm
- prior information
- gaussian process
- noise level
- sample size
- maximum likelihood
- incremental learning
- regularization parameter
- maximum a posteriori
- gaussian processes
- incomplete data
- missing values
- parameter space
- expectation maximization
- grid search
- parameter settings
- prior knowledge
- image segmentation
- probabilistic model