Complex Moment-Based Supervised Eigenmap for Dimensionality Reduction

• DOI: 10.1609/aaai.v33i01.33013910
• Publication Date: 2019-09-04T07:35:42Z
• Authors: Akira Imakura; Momo Matsuda; Xiucai Ye; Tetsuya Sakurai
• Keywords: TRACE (psycholinguistics); Univariate; Matrix (chemical analysis)

Dimensionality reduction methods that project highdimensional data to a low-dimensional space by matrix trace optimization are widely used for clustering and classification. The problem leads an eigenvalue subspace construction, preserving certain properties of the original data. However, most existing use only few eigenvectors construct space, which may lead loss useful information achieving successful Herein, overcome deficiency loss, we propose novel complex moment-based supervised eigenmap including multiple dimensionality reduction. Furthermore, proposed method provides general formulation incorporate with ridge regression, models linear dependency between covariate variables univariate labels. To reduce computational complexity, also efficient parallel implementation method. Numerical experiments indicate is competitive compared recognition performance. Additionally, exhibits high efficiency.