Author: Silas Garrido
Online AdaBoost-based M1 (OABM1) and Online AdaBoost-based M2 (OABM2) [1] are online boosting methods based on AdaBoost.M1 and AdaBoost.M2, respectively. The first changes the function responsible for updating the instances weights of OzaBoost, maintaining the characteristics of the traditional batch version. The second is targeted at multiclass problems, focusing not only on difficult instances but also on difficult classes. The corresponding codes can be downloaded via the links below:
[1] Silas G. T. C. Santos and Roberto S. M. Barros, “Online AdaBoost-based methods for multiclass problems” in Artificial Intelligence Review, 2019.
Silas Garrido - sgtcs@cin.ufpe.br