Mending is Better than Ending: Adapting Immutable Classifiers to Nonstationary Environments using Ensembles of Patches.
Sebastian KauschkeLukas FleckensteinJohannes FürnkranzPublished in: IJCNN (2019)
Keyphrases
- non stationary
- ensemble learning
- decision trees
- ensemble classifier
- classifier ensemble
- concept drift
- diversity measures
- weighted voting
- random fields
- multiple classifier systems
- support vector
- training set
- imbalanced data
- adaptive algorithms
- autoregressive
- combining classifiers
- machine learning algorithms
- training data
- classification algorithm
- naive bayes
- decision stumps
- fractional brownian motion
- ensemble members
- trained classifiers
- feature ranking
- feature selection
- empirical mode decomposition
- majority voting
- classification models
- machine learning methods
- class labels
- training samples
- feature extraction
- learning algorithm