Wuertz, R. P., Konen, W., and Behrmann, K. O. (1999). On the performance of neuronal matching algorithms. Neural Networks, 12(1):127--134. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
How Fast Can Neuronal Algorithms Match Patterns? - Würtz, Konen, Behrmann (1996) Self-citation (Konen Behrmann) (Correct)
....makes no difference. The only relevant parameter is the blob size B. Simulations in [8] have shown that convergence is fastest if a blob covers half the layer area. This blob size, which we used for all DLM simulations, can be shown to maximize the average information gain per iteration step (see [9] for details) The experimental conditions have been identical to those for the SOM. The scaling behavior, however, was different. On variation of the problem sizes N the iteration steps needed for convergence increase only linearly with N (figure 2) in sharp contrast with the exponential ....
....This figure indicates that DLM converges faster once the layer size exceeds 16 Theta 16. We conclude that the convergence time for DLM will scale as N 4 and the one for the SOM as exp(N ) Experiments that check this trend for higher values of N are currently performed and will be reported in [9]. 150 10000 100000 1e 06 1e 07 1e 08 1e 09 Fig. 3. Floating point operations required for the SOM and DLM on the mirror benchmark. The data indicate that DLM is faster for layers larger than 16 Theta 16. A question that can not be ignored in the comparison of neuronal algorithms as models for ....
R.P. Wurtz, W. Konen, and K.-O. Behrmann. On the performance of neuronal matching algorithms. Manuscript in preparation.
Bibliography of Self-Organizing Map (SOM) Papers.. - Merja Oja, Samuel.. (2002) (Correct)
No context found.
Wuertz, R. P., Konen, W., and Behrmann, K. O. (1999). On the performance of neuronal matching algorithms. Neural Networks, 12(1):127--134.
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