Fort J.C. and Pages G., Quantization versus Organization in the Kohonen S.O.I/Z, Proc. ESANN'96, M.Verleysen Ed., Editions D Facto, Bruxelles, 85-89, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the quantization aspect of the Kohonen algorithm, it is clear that minimising n,v is not equivalent to minimising , and thus that it is hopeless to expect that the Kohonen algorithm with v 1 neighbours could lead to a minimisation of distortion n. For example, we can mention the result in [12] there the difference between ,v and is measured when dim(2) 1 and centroids are arranged on a line with 2 neighbours; it is shown that the order of magnitude of min( 0 min( is C n 3, where C is a constant. In practical situations however, the learning phase of the Kohonen algorithm is ....

Fort J.C. and Pages G., Quantization versus Organization in the Kohonen S.O.I/Z, Proc. ESANN'96, M.Verleysen Ed., Editions D Facto, Bruxelles, 85-89, 1996.

....quantization aspect of the Kohonen algorithm, it is clear that minimising x n,v is not equivalent to minimising x n , and thus that it is hopeless to expect that the Kohonen algorithm with v 1 neighbours could lead to a minimisation of distortion x n . For example, we can mention the result in [12] there the difference between x n,v and x n is measured when dim(W) 1 and centroids are arranged on a line with 2 neighbours; it is shown that the order of magnitude of min(x n,v ) min(x n ) is C n 3 , where C is a constant. In practical situations however, the learning phase of the Kohonen ....

Fort J.C. and Pags G., Quantization versus Organization in the Kohonen S.O.M., Proc. ESANN'96, M.Verleysen Ed., Editions D Facto, Bruxelles, 85-89, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC