Transactions of Nanjing University of Aeronautics & Astronautics
SELFDEPENDENT LOCALITY PRESERVING PROJECTION WITH TRANSFORMED SPACEORIENTED NEIGHBORHOOD GRAPH
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Locality preserving projection (LPP) is a typical and popular dimensi onality reduction (DR) method, and it can potentially find discriminative projec tion directions by preserving the local geometric structure in data. However, LP P is based on the neighborhood graph artificially constructed from the original data , and the performance of LPP relies on how well the neares t ne ighbor criterion work in the original space. To address this issue, a novel DR algo rithm, called the selfdependent LPP (sdLPP) is proposed. And it is based on th e fact that the nearest neighbor criterion usually achieves better performance in LPP t ransformed space than that in the original space. Firstly, LPP is performed base d on the typical neighborhood graph; then, a new neighborhood graph is constructed i n LPP transformed space and repeats LPP. Furthermore, a new criterion, called the impro ved Laplacian score, is developed as an empirical reference for the discriminati v e power and the iterative termination. Finally, the feasibility and the effectiv enes s of the method are verified by several publicly available UCI and face data set s with promising results.
Keywords:
Project Supported:
Clc Number:
Mobile website
