For many clustering algorithms, it is very important to determine an appropriate number of clusters, which is called cluster validity problem. In this paper, we offer a new approach to tackle this issue. The main point is that the better outputs of clustering algorithm, the more stable. Therefore, we establish the relation between cluster validity and stability of clustering algorithms, and propose that the conditional number of Hessian matrix of the objective function with respect to outputs of the clustering algorithm can be used as cluster validity cluster index. Based on such idea, we study the traditional fuzzy c-means algorithms. Comparison experiments suggest that such a novel cluster validity index is valid for evaluating the performance of the fuzzy c-means algorithms.


Cluster Algorithm Cluster Result Stability Index Cluster Validity Stable Fixed Point 
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.


  1. 1.
    Pal, N.R., Bezdek, J.C.: On cluster validity for the fuzzy c-means model. IEEE Trans. Fuzzy Systems 3(3), 370–379 (1995)CrossRefGoogle Scholar
  2. 2.
    Bezdek, J.C.: Pattern recognition with fuzzy objective function algorithms. Plenum Press, New York (1981)zbMATHGoogle Scholar
  3. 3.
    Catherine, A., Sugar, James, G.M.: Finding the number of clusters in a dataset: an information-theoretic approach. Journal of the American Statistical Association 98(463), 750–763 (2003)zbMATHCrossRefMathSciNetGoogle Scholar
  4. 4.
    Tibshirani, R., Walther, G., Hastie, T.: Estimating the number of clusters in a data set via the gap statistic. J.R.Statist.Soc.B 63,Part 2, 411–423 (2001)zbMATHCrossRefMathSciNetGoogle Scholar
  5. 5.
    Bezdek, J.C.: Cluster validity with fuzzy sets. J.Cybernt. 3(3), 58–72 (1974)CrossRefMathSciNetGoogle Scholar
  6. 6.
    Windham, M.P.: Cluster validity for the fuzzy c-means clustering algorithm. IEEE Trans. PAMI PAMI-4(4), 357–363 (1982)Google Scholar
  7. 7.
    Windham, M.P.: Cluster validity for fuzzy clustering algorithms. Fuzzy Sets Systems 5, 177–185 (1981)zbMATHCrossRefGoogle Scholar
  8. 8.
    Backer, E., Jain, A.K.: A Cluster performance measure based on fuzzy set decomposition. IEEE Trans. PAMI PAMI-3(1) (January 1981)Google Scholar
  9. 9.
    Xie, X.L., Beni, G.: A validity measure for fuzzy clustering. IEEE Trans. PAMI 13(8), 841–847 (1991)Google Scholar
  10. 10.
    Gunderson, R.: Applications of fuzzy ISODATA algorithms to startracker printing systems. In: Proc. 7th Triannual World IFAC Congr., pp. 1319–1323 (1978)Google Scholar
  11. 11.
    Bezdek, J.C.: A physical interpretation of Fuzzy ISODATA. IEEE Trans. SMC SMC-6, 387–390 (1976)MathSciNetGoogle Scholar
  12. 12.
    Halkidi, M., Batistakis, Y., Vazirgiannis, M.: Cluster algorithms and validity measures. In: Proceedings of Thirteenth International Conference on Scientific and Statistical Database Management, pp. 3–22 (2001)Google Scholar
  13. 13.
    Jian, Y., Qiansheng, C.: The upper bound of the optimal number of clusters in fuzzy clustering. Science in China, series F 44(2), 119–125 (2001)Google Scholar
  14. 14.
    Fukuyanma, Y., Sugeno, M.: A new method of choosing the number of clusters for the fuzzy c-means method. In: Proc. 5th Fuzzy Syst. Symp., pp. 247–250 (1989) (in Japanese)Google Scholar
  15. 15.
    Wei, W., Mendel, J.M.: Optimality tests for the fuzzy c-means algorithm. Pattern Recognition 27(11), 1567–1573 (1994)zbMATHCrossRefGoogle Scholar
  16. 16.
    Jian, Y., Houkuan, H., Shengfeng, T.: An Efficient Optimality Test for the Fuzzy c- Means Algorithm. In: Proceedings of the 2002 IEEE International Conference on Fuzzy Systems, vol. 1, pp. 98–103 (2002)Google Scholar
  17. 17.
    Jian, Y.: General c-means clustering model and its applications. In: CVPR 2003, June 2003, vol. 2, pp. 122–127 (2003)Google Scholar
  18. 18.
    Anderson, E.: The IRISes of the Gaspe Peninsula. Bull. Amer. IRIS Soc. 59, 2–5 (1935)Google Scholar
  19. 19.
    Jian, Y., Qiansheng, C., Houkuan, H.: Analysis of the weighting exponent in the FCM. IEEE Transactions on Systems, Man and Cybernetics-part B: Cybernetics 34(1), 634–639 (2004)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2004

Authors and Affiliations

  • Jian Yu
    • 1
  • Houkuan Huang
    • 1
  • Shengfeng Tian
    • 1
  1. 1.Dept. of Computer ScienceBeijing Jiaotong UniversityBeijingP.R.China

Personalised recommendations