An Adaptive Heterogeneous Online Learning Ensemble Classifier for Nonstationary Environments

Algorithm 2

HDES-AD active drift handling.

	Input:: set of accurate and diverse learners
	: local prediction from base learners
	: Drift Detection Method
(1)	Drift Detection Method ()
(2)	Detect Concept Drift with DDM
(3)	If ( = true)
(4)	For learner = 1 to
(5)	reset
(6)	For learner = 1 to
(7)	Compute accuracy and diversity
(8)	End for
(9)	End if
(10)	If ( warning = true)//warning detected
(11)	For j = 1 to //loop over learners
(12)	Train new learner with new instance
(13)	End for
(14)	End if