Abstract—In order to improve the error rate estimation method, a bootstrap method was proposed. We focus on improvement of not the error rate estimation but a classification performance of a classifier. We explore a bootstrap approach for designing a classifier. In this paper, in order to improve the classification performance of a classifier, we propose a round shape bootstrap method. The areas of bootstrap samples generated by the round shape bootstrap method are expected to be more smoothed. Experimental results show the proposed method is effective for a nearest neighbor (1-NN) classifier.
Index Terms—Bootstrap samples, artificial sample generation, classifier design, 1-NN, pattern recognition, classification performance.
Yoshihiro Mitani is with the Department of Intelligent System Engineering, National Institute of Technology, Ube College, Ube, Japan (e-mail: mitani@ube-k.ac.jp).
Yusuke Fujita, and Yoshihiko Hamamoto are with Yamaguchi University, Ube, Japan.
[PDF]
Cite: Yoshihiro Mitani, Yusuke Fujita, and Yoshihiko Hamamoto, "Use of Round Shape Bootstrap Samples for a Classifier Design," International Journal of Engineering and Technology vol. 9, no. 2, pp. 179-182, 2017.
