Active learning using smooth relative regret approximations with applications.
Nir AilonRon BegleiterEsther EzraPublished in: J. Mach. Learn. Res. (2014)
Keyphrases
- active learning
- online learning
- learning strategies
- lower bound
- learning algorithm
- expert advice
- random sampling
- loss function
- training examples
- machine learning
- supervised learning
- relevance feedback
- pairwise
- experimental design
- selective sampling
- confidence bounds
- worst case
- labeled data
- semi supervised learning
- game theory
- cost sensitive
- training set
- approximation methods
- weighted majority
- loss bounds