Affiliation:
1. Jeonbuk National University
2. Jeon Buk National University
Abstract
This paper proposes a novel clustering algorithm named global kernel fuzzy-c means (GK-FCM) to segment the speech into small non-overlapping blocks for consonant/vowel segmentation. This algorithm is realized by embedding global optimization and kernelization into the classical fuzzy c-means clustering algorithm. It proceeds in an incremental way attempting to optimally add new cluster center at each stage through the kernel-based fuzzy c-means. By solving all the intermediate problems, the final near-optimal solution is determined in a deterministic way. This algorithm overcomes the well-known shortcomings of fuzzy c-means and improves the clustering accuracy. Simulation results demonstrate the effectiveness of the proposed method in consonant/vowel segmentation.
Publisher
Trans Tech Publications, Ltd.
Reference17 articles.
1. B. Balasko, J. Abonyi, and B. Feil, Fuzzy clustering and data analysis toolbox, Department of Process Engineering, University of Veszprem, Hungary, (2005).
2. J. C. Dunn, A fuzzy relative of the isodata process and its use in detecting compact well-separated clusters, J. Cybern., 3 (1974) 32-57.
3. J. C. Bezdek, Pattern recognition with fuzzy objective function algorithms, Plenum Press, New York, (1981).
4. D. Q. Zhang and S. C. Chen, Clustering incomplete data using kernel-based fuzzy c-means algorithm, Neural Process. Lett., 18 (2003) 155-162.
5. H. B. Shen, J. Yang, S. T. Wang, and X. J. Liu, Attribute weighted mercer kernel based fuzzy clustering algorithm for general non-spherical datasets, Soft Comput., 10 (2006) 1061-1073.