HyperSTAR: Task-Aware Hyperparameters for Deep Networks.
Gaurav MittalChang LiuNikolaos KarianakisVictor FragosoMei ChenYun FuPublished in: CoRR (2020)
Keyphrases
- hyperparameters
- model selection
- cross validation
- support vector
- closed form
- bayesian inference
- prior information
- random sampling
- gaussian process
- bayesian framework
- maximum likelihood
- maximum a posteriori
- em algorithm
- gaussian processes
- sample size
- incremental learning
- regularization parameter
- noise level
- parameter settings
- incomplete data
- missing values
- image reconstruction
- prior knowledge
- bayesian networks
- computer vision
- genetic algorithm
- data sets