A Novel Multi-resolution Kernel Principle Component Analysis Method
Download as PDF
DOI: 10.23977/meimie.2019.43011
Author(s)
Jianjun Wu, Weijun Gong and Zhen Shang
Corresponding Author
Jianjun Wu
ABSTRACT
Focusing on the traditional kernel principle component analysis (KPCA) can not provide principle component information under the condition of multi-resolution, this paper proposes a novel multi-resolution kernel principle component analysis (MKPCA) method, by combining KPCA with high dimensional multi-resolution analysis theory. MKPCA can explore sample data’s characters and principle component information on the different resolution. The experiments demonstrate the feasibility of the proposed method.
KEYWORDS
Multi-resolution, principle component analysis, kernel, machine learning