SmartTuning: Selecting Hyper-Parameters of a ConvNet System for Fast Training and Small Working Memory.
Xiaqing LiGuangyan ZhangWeimin ZhengPublished in: IEEE Trans. Parallel Distributed Syst. (2021)
Keyphrases
- working memory
- hyperparameters
- model selection
- cognitive load
- cross validation
- information processing
- cognitive architecture
- closed form
- computational model
- focus of attention
- bayesian inference
- noise level
- support vector
- em algorithm
- long term memory
- bayesian framework
- individual differences
- short term memory
- random sampling
- training set
- working memory capacity
- incomplete data
- prior information
- maximum likelihood
- incremental learning
- expectation maximization
- parameter space
- prior knowledge