Combining similarity in time and space for training set formation under concept drift.
Indre ZliobaitePublished in: Intell. Data Anal. (2011)
Keyphrases
- concept drift
- training set
- classification algorithm
- data streams
- streaming data
- class distribution
- drift detection
- drifting concepts
- non stationary
- classification accuracy
- incremental clustering
- evolving data streams
- data stream mining
- data stream classification
- nearest neighbor
- training data
- batch learning
- change detection
- k nearest neighbor
- training samples
- active learning
- decision trees
- low dimensional
- data distribution
- supervised learning
- management system
- similarity measure
- pattern recognition
- data sets
- image segmentation
- feature space
- concept descriptions
- pairwise
- class imbalance
- data management
- sliding window
- database