Dimensionality reduction of image feature based on geometric parameter adaptive LLE algorithm

Issue title: Special issue: Fuzzy Systems in Distributed Sensing Applications

Guest editors: Mohamed Elhoseny and X. Yuan

Article type: Research Article

Authors: Fan, Linyuan^{; *}

Affiliations: School of Statistics, Capital University of Economics and Business, Beijing, China

Correspondence: [*] Corresponding author. Linyuan Fan, School of Statistics, Capital University of Economics and Business, Beijing, 100070, China. E-mail: fanlinyuan@cueb.edu.cn.

Abstract: Locally linear embedding (LLE) is a classical nonlinear dimensionality reduction algorithm, and it has been widely used in image feature selection. LLE reduces the dimensions of a data set only by exploring the geometric structure, which is calculated by Euclidean distance and makes the embedding result be sensitive to noise. Moreover, the choice of the number of nearest neighbors is fixed for all data points and only given by human experience. In order to overcome these problems, a geometric parameter adaptive LLE (PALLE) algorithm is proposed in this paper. This algorithm jointly uses Geodesic distance and Cosine similarity to replace Euclidean distance, and then the number of neighbors is adaptable selected by weak-σ rule. Extensive experimental results over various real-life data sets have demonstrated the superiority of the proposed algorithm in terms of image feature dimensionality reduction compared with classical LLE and other well-known algorithms.

Keywords: Image feature, dimensionality reduction, LLE, parameter adaptive

DOI: 10.3233/JIFS-179520

Journal: Journal of Intelligent & Fuzzy Systems, vol. 38, no. 2, pp. 1569-1577, 2020