Advertisement

A New Principal Curve Algorithm for Nonlinear Principal Component Analysis

  • David Antory
  • Uwe Kruger
  • Tim Littler
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4113)

Abstract

This paper summarizes a new concept to determine principal curves for nonlinear principal component analysis (PCA). The concept is explained within the framework of the Hastie and Stuetzle algorithm and utilizes spline functions. The paper proposes a new algorithm and shows that it provides an efficient method to extract underlying information from measured data. The new method is geometrically simple and computationally expedient, as the number of unknown parameters increases linearly with the analyzed variable set. The utility of the algorithm is exemplified in two examples.

Keywords

Principal Curve Spline Function Projection Stage Nonlinear Principal Component Analysis Spline Parameter 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Delicado, P.: Another Look at Principal Curves and Surfaces. Commun. Stat. Part. B Simul. Comput. 77, 84–116 (2001)zbMATHMathSciNetGoogle Scholar
  2. 2.
    Dong, D., McAvoy, T.J.: Non-Linear Principal Component Analysis – Based on Principal Curves and Neural Networks. Comp. Chem. Engng. 20(1), 65–78 (1996)CrossRefGoogle Scholar
  3. 3.
    Hastie, T., Stuetzle, W.: Principal Curves. J. Am. Stat. Assoc. 84(406), 502–517 (1989)zbMATHCrossRefMathSciNetGoogle Scholar
  4. 4.
    Jia, F., Martin, E.B., Morris, A.J.: Non-Linear Principal Component Analysis with Applications to Process Fault Detection. Int. J. Syst. Sci. 31(11), 1473–1487 (2000)zbMATHCrossRefGoogle Scholar
  5. 5.
    Kegl, B., Krzyzak, A., Linder, T., Zeger, K.: Learning and Design of Principal Curves. IEEE Trans. Pattern. Anal. Mach. Intell. 22(3), 281–297 (2000)CrossRefGoogle Scholar
  6. 6.
    Kramer, M.A.: Non-linear Principal Component Analysis using Auto-Associative Neural Networks. AIChE J. 37(2), 233–243 (1991)CrossRefGoogle Scholar
  7. 7.
    MacGregor, J.F., Marlin, T.E., Kresta, J.V., Skagerberg, B.: Multivariate Statistical Methods in Process Analysis and Control. In: Proceedings of the 4th International Conference on Chemical Process Control, pp. 79–99. AIChE Publ. No. P-67, New York (1991)Google Scholar
  8. 8.
    Qin, S.J., McAvoy, T.J.: Nonlinear PLS Modelling Using Neural Networks. Comp. Chem. Engng. 16(4), 379–391 (1992)CrossRefGoogle Scholar
  9. 9.
    Sharma, S.K., Irwin, G.W.: Fuzzy Coding of Genetic Algorithms. IEEE Trans. Evol. Comput. 7(4), 344–355 (2003)CrossRefGoogle Scholar
  10. 10.
    Tibshirani, R.: Principal Curves Revisited. Stat. Comput. 2, 183–190 (1992)CrossRefGoogle Scholar
  11. 11.
    Wilson, D.J.H., Irwin, G.W., Lightbody, G.: RBF Manifolds for Process Monitoring. IEEE Trans. Neural. Netw. 10(6), 1424–1434 (1999)CrossRefGoogle Scholar
  12. 12.
    Wise, B.M., Gallagher, N.B.: The Process Chemometrics Approach to Process Monitoring and Fault Detection. J. Process. Control 6(6), 329–348 (1996)CrossRefGoogle Scholar
  13. 13.
    Wold, S.: Spline functions in data analysis. Technometrics 16, 1–11 (1974)zbMATHCrossRefMathSciNetGoogle Scholar
  14. 14.
    Zhang, F.: Identifying Nonlinear Variation Patterns in Multivariate Manufacturing Processes, Ph.D. Thesis, Texas A&M University (2004)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • David Antory
    • 1
  • Uwe Kruger
    • 2
  • Tim Littler
    • 3
  1. 1.International Automotive Research CentreUniversity of WarwickCoventryU.K.
  2. 2.Intelligent Systems and Control GroupQueen’s UniversityBelfastU.K.
  3. 3.Energy Systems Research GroupQueen’s UniversityBelfastU.K.

Personalised recommendations