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