An effective principal curves extraction algorithm for complex distribution dataset

  • Authors:

  • Hongyun Zhang;Duoqian Miao;Lijun Sun;Ying Ye

  • Affiliations:

  • The Key Laboratory of Embedded System and Service Computing, Ministry of Education, China, Tongji University, Shanghai, China and School of Electronic and Information Engineering, Tongji Universit ...;The Key Laboratory of Embedded System and Service Computing, Ministry of Education, China, Tongji University, Shanghai, China and School of Electronic and Information Engineering, Tongji Universit ...;The Key Laboratory of Embedded System and Service Computing, Ministry of Education, China, Tongji University, Shanghai, China and School of Electronic and Information Engineering, Tongji Universit ...;The Key Laboratory of Embedded System and Service Computing, Ministry of Education, China, Tongji University, Shanghai, China and School of Electronic and Information Engineering, Tongji Universit ...

  • Venue:
  • RSKT'10 Proceedings of the 5th international conference on Rough set and knowledge technology
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a new method for finding principal curves from complex distribution dataset. Motivated by solving the problem, which is that existing methods did not perform well on finding principal curve in complex distribution dataset with high curvature, high dispersion and self-intersecting, such as spiral-shaped curves, Firstly, rudimentary principal graph of data set is created based on the thinning algorithm, and then the contiguous vertices are merged. Finally the fitting-and-smoothing step introduced by Kégl is improved to optimize the principal graph, and Kégl's restructuring step is used to rectify imperfections of principal graph. Experimental results indicate the effectiveness of the proposed method on finding principal curves in complex distribution dataset.