MINAS: multiclass learning algorithm for novelty detection in data streams.
Elaine Ribeiro de FariaAndré Carlos Ponce de Leon Ferreira de CarvalhoJoão GamaPublished in: Data Min. Knowl. Discov. (2016)
Keyphrases
- novelty detection
- multi class
- data streams
- concept drift
- learning algorithm
- perceptron algorithm
- classification algorithm
- support vector machine
- boosting algorithms
- multiclass problems
- active learning
- anomaly detection
- binary classification
- training data
- multiclass classification
- learning problems
- supervised learning
- mistake bound
- cost sensitive
- change detection
- machine learning algorithms
- multiclass support vector machines
- multiple classes
- non stationary
- pairwise
- multiclass learning
- reinforcement learning
- feature selection
- data sets
- machine learning
- multi task
- generalization ability
- learning tasks
- data distribution
- data structure
- generalization error
- binary classifiers
- learning process
- error correcting output codes
- image processing
- multi task learning
- transfer learning
- training samples
- labeled data