Critical Bach Size Minimizes Stochastic First-Order Oracle Complexity of Deep Learning Optimizer using Hyperparameters Close to One.
Hideaki IidukaPublished in: CoRR (2022)
Keyphrases
- deep learning
- hyperparameters
- model selection
- closed form
- unsupervised learning
- cross validation
- bayesian framework
- gaussian process
- machine learning
- support vector
- sample size
- bayesian inference
- em algorithm
- prior information
- incremental learning
- random sampling
- higher order
- missing values
- training data
- incomplete data
- data mining
- weakly supervised
- pattern recognition
- maximum a posteriori
- data analysis
- image reconstruction
- prior knowledge