Role of Hyperparameters in Deep Active Learning.
Denis HuseljicMarek HerdePaul HahnBernhard SickPublished in: IAL@PKDD/ECML (2023)
Keyphrases
- hyperparameters
- active learning
- random sampling
- model selection
- cross validation
- bayesian inference
- gaussian process
- closed form
- em algorithm
- bayesian framework
- prior information
- support vector
- maximum likelihood
- maximum a posteriori
- sample size
- incremental learning
- regularization parameter
- semi supervised
- training set
- machine learning
- parameter settings
- incomplete data
- prior knowledge
- e learning
- learning algorithm
- grid search
- parameter space
- sliding window
- error rate
- supervised learning
- probabilistic model
- learning process
- data streams