On the convergence of loss and uncertainty-based active learning algorithms.
Daniel HaimovichDima KaramshukFridolin LinderNiek TaxMilan VojnovicPublished in: CoRR (2023)
Keyphrases
- learning algorithm
- iterative algorithms
- machine learning
- back propagation
- training data
- convergence speed
- supervised learning
- convergence rate
- machine learning algorithms
- learning tasks
- uncertain data
- training examples
- learning scheme
- reinforcement learning
- global convergence
- batch mode
- decision tree learning
- general loss functions
- missing data
- conditional probabilities
- training samples
- active learning
- restricted boltzmann machine