Paper:
Clustering Algorithm Based on Probabilistic Dissimilarity
Makito Yamashiro*, Yasunori Endo**, and Yukihiro Hamasuna*
*Graduate School of Systems and Information Engineering, University of Tsukuba
**Department of Risk Engineering, Faculty of Systems and Information Engineering, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8573, Japan
The clustering algorithm we propose is based on probabilistic dissimilarity, which is formed by introducing the concept of probability into conventional dissimilarity. After defining probabilistic dissimilarity, we present examples of probabilistic dissimilarity functions. After considering an objective function with probabilistic dissimilarity. Furthermore, we construct a clustering algorithm probabilistic dissimilarity based using optimal solutions maximizing the objective function. Numerical examples verify the effectiveness of our algorithm.
- [1] J. C. Bezdek, “Pattern Recognition with Fuzzy Objective Function Algorithms,” Plenum, New York, 1991.
- [2]K. Jajuga, “Classification, Clustering, and Data Analysis: Recent Advances and Applications,” Springer, 2002.
- [3] B. Schweizer and A. Solar, “Probabilistic Metric Spaces,” Dover Publications, Inc. Mineola, New York, 2005.
- [4] O. Hadzic and A. Pap, “Fixed Point Theory in Probabilistic metric Spaces,” KLUWER ACADEMIC PUBLISHERS, 2001.
- [5] K. Menger, “Statistical metrics,” Proc. Nat Acad. of Sci., U.S.A., 28, pp. 535-537, 1942.