Improving RBF Networks Classification Performance by using K-Harmonic Means
In this paper, a clustering algorithm named KHarmonic means (KHM) was employed in the training of Radial Basis Function Networks (RBFNs). KHM organized the data in clusters and determined the centres of the basis function. The popular clustering algorithms, namely K-means (KM) and Fuzzy c-means (FCM), are highly dependent on the initial identification of elements that represent the cluster well. In KHM, the problem can be avoided. This leads to improvement in the classification performance when compared to other clustering algorithms. A comparison of the classification accuracy was performed between KM, FCM and KHM. The classification performance is based on the benchmark data sets: Iris Plant, Diabetes and Breast Cancer. RBFN training with the KHM algorithm shows better accuracy in classification problem.
- C.M. Bishop, Neural Networks for Pattern Recognition. Oxford University Pres, New York, USA, 1995.
- N. Benoudjit, M. Verleysen, On the kernel width in radial basis function networks. Neural Processing Letters 18, 2003, 139-154.
- J. Moody, C.J. Darken, Fast learning in networks of locally-tuned processing unit. Neural Computation 1, 1989, 281-294.
- K.Warwick, J.D. Mason and E.L. Sutanto, Neural network basis function center selection using cluster analysis. Proceeding of American Central Conference, Washington, June, 1995
- J.C. Dunn, A fuzzy relative of the ISODATA process and its use n detecting compact well-separated clusters. J. Cybernet. 3, 1973, 32-57.
- J. Bezdek, Pattern recognition with fuzzy objective function algorithm, Plenum Pres, NewYork, 1981.
- R.L. Canon, J.Dave and J.C. Bezdek, Efficient implementation of the fuzzy cmeans clustering algorithms. IEEE Trans Pattern Arial Machine, Intell 8, 248-255.
- B. Zhang, M. Hsu, U. Dayal, K-harmonic means - a data clustering algorithm, Technical Report HPL-1999-124, Hewlett -Packard Laboratories, 1999.
- B. Zhang, M. Hsu, U. Dayal, K-harmonic means, in: International Workshop on Temporal, Spatial and Spatio-Temporal Data Mining, TSDM2000, Lyon, France, 12 September 2000.  G. Hammerly, C. Elken, Alternatives to the K-means algorithm that find better clusterins, in: Proceedings of the 11th International Conference on Information and Knowledge Management, 2002, pp. 600-607.  C.L. Blake, C.J. Merz, UCI repository of machine learning databases, 2008, http://archive.ics.uci.edu/ml/databases.html.