Rethinking the Hyperparameters for Fine-tuning.
Hao LiPratik ChaudhariHao YangMichael LamAvinash RavichandranRahul BhotikaStefano SoattoPublished in: ICLR (2020)
Keyphrases
- fine tuning
- hyperparameters
- model selection
- cross validation
- closed form
- bayesian inference
- support vector
- random sampling
- bayesian framework
- prior information
- gaussian process
- em algorithm
- sample size
- noise level
- incremental learning
- maximum likelihood
- gaussian processes
- maximum a posteriori
- incomplete data
- regularization parameter
- missing values
- parameter space
- fine tuned
- data sets
- grid search
- parameter values
- expectation maximization
- feature space
- data streams
- multiscale
- image processing
- genetic algorithm
- data mining