A new globally adaptive k -nearest neighbor classifier based on local mean optimization
- PDF / 1,508,450 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 35 Downloads / 197 Views
(0123456789().,-volV)(0123456789().,-volV)
METHODOLOGIES AND APPLICATION
A new globally adaptive k-nearest neighbor classifier based on local mean optimization Zhibin Pan1,2 • Yiwei Pan1 • Yidi Wang1 • Wei Wang3
Ó Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The k-nearest neighbor (KNN) rule is a simple and effective nonparametric classification algorithm in pattern classification. However, it suffers from several problems such as sensitivity to outliers and inaccurate classification decision rule. Thus, a local mean-based k-nearest neighbor classifier (LMKNN) was proposed to address these problems, which assigns the query sample with a class label based on the closest local mean vector among all classes. It is proven that the LMKNN classifier achieves better classification performance and is more robust to outliers than the classical KNN classifier. Nonetheless, the unreliable nearest neighbor selection rule and single local mean vector strategy in LMKNN classifier severely have negative effect on its classification performance. Considering these problems in LMKNN, we propose a globally adaptive k-nearest neighbor classifier based on local mean optimization, which utilizes the globally adaptive nearest neighbor selection strategy and the implementation of local mean optimization to obtain more convincing and reliable local mean vectors. The corresponding experimental results conducted on twenty real-world datasets demonstrated that the proposed classifier achieves better classification performance and is less sensitive to the neighborhood size k compared with other improved KNN-based classification methods. Keywords k-nearest neighbors Pattern classification Globally adaptive nearest neighbors Local mean optimization
1 Introduction As one of the most famous top 10 algorithms in data mining, k-nearest neighbor (KNN) rule (Cover and Hart 1967) has been deeply studied and widely applied in the field of pattern classification. The KNN classifier is a wellknown classification method in the field of nonparametric classifiers due to its simplicity, effectiveness and intuitiveness (Jiang et al. 2012). To be specific, the query
Communicated by V. Loia. & Zhibin Pan [email protected] 1
School of Electronic and Information Engineering, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China
2
National Key Laboratory of Science and Technology on Space Microwave, CAST, Xi’an, People’s Republic of China
3
National Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences (CASIA), Beijing, People’s Republic of China
sample is assigned to the class through a majority vote strategy in its k-nearest neighborhood. Compared with other classifiers such as decision tree, Bayes classifier, SVM, neural network and so on, KNN is a nonparametric classification algorithm and do not need any training or learning process. In detail, KNN classifier does not require any prior knowledge about parametric information on training samples (Li et al. 2008)
Data Loading...
