Relevance and Kernel Self-Organising Maps
We review the recently proposed method of Kernel Self-organising maps (KSOM) which has been shown to exhibit very fast convergence. We show that this is due to an interaction between the fact that we are working in an overcomplete basis and the fact that we are using a mixture of one-shot and incremental learning. We then review Relevance Vector Machines which is a supervised training method related to Support Vector Machines and apply it to creating the Relevance Self-Organising Map, RSOM. We show results on artificial data and on the iris data set.
KeywordsSparse Representation Neighbourhood Function Relevance Vector Machine Kernel Space Winning Neuron
Unable to display preview. Download preview PDF.
- 1.D. MacDonald and C. Fyfe. The kernel self-organising map. In R.J. Howlett and L. C. Jain, editors, Fourth International Conference on Knowledge-based Intelligent Engineering Systems and Allied Technologies, KES2000, 2000.Google Scholar
- 2.D. Michie, D. J. Spiegelhalter, and C. C. Taylor, editors. Machine learning, neural and statistical classification. Ellis Horwood, 1994.Google Scholar
- 3.B. Schölkopf. The kernel trick for distances. Technical report, Microsoft Research, May 2000.Google Scholar
- 5.A. J. Smola, S. Mika, B. Schölkopf, and R. C. Williamson. Regularized principal maniforlds. Machine Learning, pages 1–28, 2000.Google Scholar
- 6.M. Tipping. The relevance vector machine. In S. A. Solla, T. K. Leen, and K.-R. Muller, editors, Advances in Neural Information Processing Systems, 12. MIT Press, 2000.Google Scholar