Heuristic improvement for active learning using localized generalization error as selection criterion.
Wing W. Y. NgBinbin SunDaniel S. YeungXizhao WangPublished in: SMC (2007)
Keyphrases
- generalization error
- selection criterion
- active learning
- model selection
- learning algorithm
- cross validation
- query by committee
- binary classification
- classification error
- sample complexity
- variable selection
- mutual information
- feature selection
- machine learning
- optimal solution
- sample size
- leave one out cross validation
- training error
- learning machines
- training set
- supervised learning
- random sampling
- support vector machine
- learning process
- objective function
- decision trees
- information criterion
- semi supervised
- feature selection algorithms
- search algorithm
- simulated annealing
- neural network
- labeled data
- boosting algorithms
- data sets
- upper bound
- genetic algorithm