Rethinking Binary Hyperparameters for Deep Transfer Learning.
Jo PlestedXuyang ShenTom GedeonPublished in: ICONIP (2) (2021)
Keyphrases
- transfer learning
- hyperparameters
- model selection
- cross validation
- support vector
- random sampling
- bayesian framework
- learning tasks
- closed form
- bayesian inference
- prior information
- active learning
- em algorithm
- sample size
- gaussian process
- knowledge transfer
- maximum a posteriori
- incremental learning
- machine learning
- reinforcement learning
- cross domain
- noise level
- labeled data
- gaussian processes
- maximum likelihood
- incomplete data
- multi task learning
- collaborative filtering
- machine learning algorithms
- missing values
- multi task
- semi supervised learning
- learning algorithm
- text classification
- text categorization
- transfer knowledge
- parameter space
- data mining
- target domain
- high resolution
- probabilistic model
- supervised learning
- expectation maximization