Negative Results for Active Learning with Convex Losses.
Steve HannekeLiu YangPublished in: AISTATS (2010)
Keyphrases
- active learning
- positive and negative
- expected loss
- semi supervised
- convex optimization
- supervised learning
- selective sampling
- learning algorithm
- random sampling
- machine learning
- learning strategies
- convex hull
- cost sensitive
- convex sets
- experimental design
- unlabeled data
- training examples
- relevance feedback
- multi class
- music emotion classification
- pool based active learning
- digital convexity
- annotation effort
- crowd sourced
- learning process
- sample selection
- piecewise linear
- constrained minimization
- convex functions
- globally optimal
- labeled data
- batch mode active learning
- semi supervised learning