Hyperbolic Quotient Feature Map for Competitive Learning Neural Networks
In this paper, we present a new learning method called hyperbolic quotient feature map for competitive learning neural networks. The previous neural network learning algorithms didn’t consider their topological properties, and thus their dynamics were not clearly defined. We show that the weight vectors obtained by competitive learning decompose the input vector space and map it to the quotient space X/R. In addition, we define a quotient function which maps [1, ∞) ⊂ R n to [0,1) and induce the proposed algorithm from the performance measure with the quotient function. Experimental results for pattern recognition of remote sensing data indicate the superiority of the proposed algorithm in comparison to the conventional competitive learning methods.
KeywordsWeight Vector Input Vector Canonical Function Quotient Function Topological Information
Unable to display preview. Download preview PDF.
- 1.Cox, D.R., Hinskley, D.V., Reid, N., Rubin, D.B., Silverman, B.W.: Networks and Chaos - Statistical and Probabilitic Aspects. Chapman & Hall, New York (1993)Google Scholar
- 2.Yin, F., Wang, J.: Advances in Neural Networks. Springer, Heidelberg (2005)Google Scholar
- 5.Gihman, I.I., Skorohod, A.V.: The Theory of Stochastic Process, 2nd edn. Springer, Heidelberg (1980)Google Scholar
- 6.Luenberger, D.G.: Optimization by vector space methods. John-Wiley and Sons, New Jersey (1968)Google Scholar