High density-focused uncertainty sampling for active learning over evolving stream data.
Dino IencoIndre ZliobaiteBernhard PfahringerPublished in: BigMine (2014)
Keyphrases
- high density
- uncertainty sampling
- active learning
- stream data
- data streams
- sliding window
- experimental design
- random sampling
- cost sensitive
- low density
- query by committee
- historical data
- concept drift
- streaming data
- data center
- ensemble members
- network traffic
- sequential patterns
- learning algorithm
- supervised learning
- sensor data
- training set
- training examples
- machine learning
- labeled data
- continuous queries
- semi supervised
- generalization error
- class imbalance
- learning process
- unlabeled data
- misclassification costs
- data distribution
- ensemble methods
- semi supervised learning
- pairwise
- naive bayes
- probability estimates
- feature selection
- data sets
- data mining
- databases