Browse > Article
http://dx.doi.org/10.5391/JKIIS.2003.13.4.416

Improved Expectation and Maximization via a New Method for Initial Values  

Kim, Sung-Soo (충북대학교 전기전자 및 컴퓨터 공학부)
Kang, Jee-Hye (충북대학교 전기전자 및 컴퓨터 공학부)
Publication Information
Journal of the Korean Institute of Intelligent Systems / v.13, no.4, 2003 , pp. 416-426 More about this Journal
Abstract
In this paper we propose a new method for choosing the initial values of Expectation-Maximization(EM) algorithm that has been used in various applications for clustering. Conventionally, the initial values were chosen randomly, which sometimes yields undesired local convergence. Later, K-means clustering method was employed to choose better initial values, which is currently widely used. However the method using K-means still has the same problem of converging to local points. In order to resolve this problem, a new method of initializing values for the EM process. The proposed method not only strengthens the characteristics of EM such that the number of iteration is reduced in great amount but also removes the possibility of falling into local convergence.
Keywords
Expectation-Maximization; K-means; Uniform Partitioning; Initial Values;
Citations & Related Records
연도 인용수 순위
  • Reference
1 H. Takagi and M. Sugeno, "Fuzzy Identification of Systems and Its Application to Modeling and Control," IEEE Trans. on Sys. Man and Cybern., Vol. 15, pp. 116-132, 1985.
2 S. Zabin and H. Poor, "Efficient estimation of class- A noise parameters via the EM algorithm," IEEE Trans. Info T., vol. 37, no. 1, pp. 60-72, 1991.   DOI   ScienceOn
3 C. Wu, "On the convergence properties of the EM algorithm," Ann. Statist., vol. 11. 1, pp. 95-103, 1983.   DOI   ScienceOn
4 H. Chen, R. Perry, and K. Buckley, "Direct and EM-based map sequence estimation with unknown time-varying channels," Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, vol. 4, pp. 2129-2132, 2001.
5 R. A. Boyles, "On the convergence of the EM algorithm," J. Roy. Sta B., vol. 45, no. 1, pp. 47-50, 1983.
6 R. J. Kozick, B. M. Sadler, "Maximum-likelihood array processing in non -Gaussian noise with Gaussian mixtures," IEEE Trans. on Signal Processing, vol. 48, No. 12, pp. 3520-3535, 2000..   DOI   ScienceOn
7 A. P. Dempster, N.M. Laird, and D.B. Rubin, "Maximum likelihood from incomplete data via the EM algorithm," J. Royal Statisticla Soc., Set. B, vol. 39, no. 1, pp.1-38, 1977.
8 For an extensive list of references to papers describing applications of the EM-algorithm, see, http://www. engineering/usu. edu/Departments/ece/Publications/Moon
9 K. K. Paliwal and V. Ramasubramainan, "Modified K-means algorithm for vector quantizer design," IEEE Image Processing Trans, vol. 9 pp. 1964-1967, Nov 2000   DOI   ScienceOn
10 R. Redner and H. F. Walker, "Mixture densities, maximum-likelihood estimation and the EM algorithm (review)," SIAM Rev., vol. 26, no. 2, pp. 195-237, 1984.   DOI   ScienceOn
11 J. H. Holland, "Genetic Algorithms in Search, Optimization and Machine Learning," Addison-Wesley, Reading, MA, 1989.
12 Wong Ching-Chang and Chen Chia-Chong, "K-means-based fuzzy classifier design," IEEE Fuzzy Systems International Conference, vol. 1, pp. 48-52, May 2000