Diverse Adaboost Through Heterogenous Component Classifiers

Citations Over Time

Abstract

Homogenous component classifiers in the Adaboost algorithm decrease the diversity among component classifiers, which degrades ensemble performance for classification. Aiming to increase the diversity, we propose heterogenous component classifiers for the Adaboost algorithm (termed as He-Adaboost). During each round, three diverse models based on totally different principles are used for learning from the distribution, and three hypotheses are obtained. The diversity of each hypothesis is measured, and the hypothesis with the maximum diversity is selected as the component classifier in this round. To evaluate the ensemble performance of the proposed He-Adaboost method, some experiments are carried out on the standard datasets. Experimental results consistently confirm the superiority of proposed method. It indicates that this study provides an encouraging methodology for improving ensemble performance of the Adaboost algorithm.

Related Papers

• → Boosting algorithms for network intrusion detection: A comparative evaluation of Real AdaBoost, Gentle AdaBoost and Modest AdaBoost(2020)204 cited
• → Advance and Prospects of AdaBoost Algorithm(2014)197 cited
• → Supplemental Boosting and Cascaded ConvNet Based Transfer Learning Structure for Fast Traffic Sign Detection in Unknown Application Scenes(2018)11 cited
• The Typical Algorithm of AdaBoost Series in Boosting Family(2003)
• → Boosting Ensembles of Weak Classifiers in High Dimensional Input Spaces(2009)