Mostafa Mohammadpour - Department of Computer Engineering, Qazvin Branch, Islamic Azad University,Qazvin, Iran
MohammadKazem Ghorbanian - Department of Computer Engineering, Buinzahra Branch, Islamic Azad University, Buinzahra, Iran
Saeed Mozaffari - Electrical and Computer Engineering Department, Semnan University, Semnan, Iran

An improved AdaBoost algorithm based on optimizingsearch in sample space is presented. Working with datain large scale need more time to compare samples for findinga threshold in the AdaBoost algorithm when using decisionstump as a weak classifier. We used PSO algorithm to evolveand select best feature in sample space for a weak classifier toreduce time. The experiment results show that with applyingPSO to the decision stump, time consuming of the AdaBoostalgorithm has been improved than base Adaboost. As a result,using evolutionary algorithms in such problems which have largescale, can reduce searching time for finding best solution andincrease performance of algorithms in hand.

AdaBoost, Evolutionary Algorithms, PSO, Decision Stump