Improved self-splitting competitive learning algorithm

Authors:
Jun Liu;Kotagiri Ramamohanarao
Affiliations:
Department of Computer Science and Software Engineering, The University of Melbourne, Victoria, Australia;Department of Computer Science and Software Engineering, The University of Melbourne, Victoria, Australia
Venue:
PAKDD'05 Proceedings of the 9th Pacific-Asia conference on Advances in Knowledge Discovery and Data Mining
Year:
2005

Citing 3
Cited 0

X-means: Extending K-means with Efficient Estimation of the Number of Clusters

ICML '00 Proceedings of the Seventeenth International Conference on Machine Learning
Review:

Neural Computation
Self-splitting competitive learning: a new on-line clustering paradigm

IEEE Transactions on Neural Networks

Quantified Score

Hi-index	0.00

Visualization

Abstract

The Self-Splitting Competitive Learning (SSCL) is a powerful algorithm that solves the difficult problems of determining the number of clusters and the sensitivity to prototype initialization in clustering. The SSCL algorithm iteratively partitions the data space into natural clusters without a prior information on the number of clusters. However, SSCL suffers from two major disadvantages: it does not have a proven convergence and the speed of learning process is slow. We propose solutions for these two problems. Firstly, we introduce a new update scheme and lead a proven convergence of Asymptotic Property Vector. Secondly, we modify the split-validity to accelerate the learning process. Experiments show these techniques make the algorithm faster than the original one.