moa.classifiers.active

Class ActiveClassifier

java.lang.Object

moa.AbstractMOAObject

moa.options.AbstractOptionHandler

moa.classifiers.AbstractClassifier

moa.classifiers.active.ActiveClassifier

All Implemented Interfaces:

Serializable, Classifier, AWTRenderable, MOAObject, OptionHandler

public class ActiveClassifier
extends AbstractClassifier

Active learning setting for evolving data streams.

Active learning focuses on learning an accurate model with as few labels as possible. Streaming data poses additional challenges for active learning, since the data distribution may change over time (concept drift) and classifiers need to adapt. Conventional active learning strategies concentrate on querying the most uncertain instances, which are typically concentrated around the decision boundary. If changes do not occur close to the boundary, they will be missed and classifiers will fail to adapt. This class contains four active learning strategies for streaming data that explicitly handle concept drift. They are based on randomization, fixed uncertainty, dynamic allocation of labeling efforts over time and randomization of the search space [ZBPH]. It also contains the Selective Sampling strategy, which is adapted from [CGZ] it uses a variable labeling threshold.

[ZBPH] Indre Zliobaite, Albert Bifet, Bernhard Pfahringer, Geoff Holmes: Active Learning with Evolving Streaming Data. ECML/PKDD (3) 2011: 597-612

[CGZ] N. Cesa-Bianchi, C. Gentile, and L. Zaniboni. Worst-case analysis of selective sampling for linear classification. J. Mach. Learn. Res. (7) 2006: 1205-1230

.

Parameters:

• -l : Classifier to train
• -d : Strategy to use: Random, FixedUncertainty, VarUncertainty, RandVarUncertainty, SelSampling

-b : Budget to use -u : Fixed threshold -s : Floating budget step -n : Number of instances at beginning without active learning

Version:

$Revision: 7 $

Author:

Indre Zliobaite (zliobaite at gmail dot com), Albert Bifet (abifet at cs dot waikato dot ac dot nz)

See Also:

Serialized Form