Year: 2022
Type: article
Abstract: Adaptive Boosting (AdaBoost) is a distinguished representative of ensemble learning algorithm which combines weak classifiers as a strong classifier by the weighted majority voting rule. Weak classifiers of AdaBoost with Threshold Classification, trying to find the best threshold in one of the data dimensions to separate the data into two classes -1 and 1, are proved to be both convenient and efficient. However, in some case, this Weak Learn algorithm achieves not a great enough accuracy and poor generalization performance due to the following reasons: (1) In each iteration, it finds threshold in all data dimensions, but only the best one with the greatest accuracy will be referenced for classification that doesn’t take advantage of the potential value of other dimensions. (2) This classifier still has the tendency of overfitting as it may focus on training examples which is misclassified by just only one criterion in each iteration. To improve its accuracy and generalization, we proposed a new Weak Learn algorithm which classifies examples based on multiple thresholds not the only one and correspondingly, the weight assigning scheme of the Weak Learn algorithm is changed for AdaBoost algorithm in this paper, theoretical identification is provided to show the superiority. Experimental studies are also presented to verify the effectiveness of the method.
Source: Applied Sciences
DOI: 10.3390/app12125872
Authors: Yi Ding, Hongyang Zhu, Ruyun Chen, Ronghui Li
Institution: Guangdong Ocean University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,408
Funder: Guangdong Ocean University
OpenAlex ID: W4281714913