DELA: A Dynamic Online Ensemble Learning Algorithm
Hamid Bouchachia, Emili Balaguer‐Ballester · The European Symposium on Artificial Neural Networks · 2014
The present paper investigates the problem of prediction in the context of dynamically changing environment, where data arrive over time. A Dynamic on- line Ensemble Learning Algorithm (DELA) is introduced. The adaptivity concerns three levels: structural adaptivity, combination adaptivity and model adaptivity. In particular, the structure of the ensemble is sought to evolve in order to be able to deal with the problem of data drift. The proposed online ensemble is evaluated on the stagger data set to show its predictive power in presence of data drift.