PBIL for Optimizing Hyperparameters of Convolutional Neural Networks and STL Decomposition.
Roberto A. Vasco-CarofilisMiguel A. Gutiérrez-NaranjoMiguel Cárdenas-MontesPublished in: HAIS (2020)
Keyphrases
- hyperparameters
- convolutional neural networks
- model selection
- closed form
- bayesian inference
- cross validation
- bayesian framework
- random sampling
- prior information
- support vector
- gaussian process
- noise level
- sample size
- em algorithm
- maximum likelihood
- convolutional network
- maximum a posteriori
- incremental learning
- evolutionary algorithm
- incomplete data
- gaussian processes
- missing values
- generative model
- parameter space
- parameter settings
- prior knowledge
- grid search
- noisy images
- image reconstruction
- probabilistic model
- multiscale
- learning algorithm