In this paper, we present a new modification of AdaBoost (Adaptive-Boosting) algorithm that improves its efficiency in presence of class-label noise. The proposed algorithm is significantly more robust against label noise than RobusBoost, the best available and implemented solution. Moreover, unlike RobusBoost, the proposed method does not need optimization toolbox or external parameters provided by user. The empirical results on 9 benchmark datasets showed the proposed method significantly outperforms RobustBoost in terms of classification accuracy and computation time in majority of the examined data sets.