Individual Classifiers Accuracy Based Online Boosting Algorithm

Boosting is a method for transforming weak learners into a powerful ensemble. To model this technique, classifiers are trained on several ‎datasets. This article proposes a boosting-like online learning algorithm, a modified form of Oza and Russell's online boosting, aimed at ‎maintaining high and consistent accuracy in environments where concepts change continuously. We use each classifier's accuracy to trans-‎form weak learners into strong learners. Our algorithm was tested against other state-of-the-art methods using several real and artificial da-‎tasets. Accuracy improved drastically in most tested situations‎.