This page is an archival copy for indexing and accessibility purposes. All figures have been removed to decrease the page size. See https://doi.org/10.1109/ijcnn.2018.8489044 for the final version.

The impact of Interconnecting Topologies on SOM Neural Networks

Mayra Z. Pimenta, Cesar Henrique Comin (ORCID), Francisco A. Rodrigues, Luciano Da F. Costa

2018 International Joint Conference on Neural Networks (IJCNN), 1: 1-6. . DOI: 10.1109/ijcnn.2018.8489044.

Abstract

The effect of different interconnecting topologies on the performance of SOM networks is investigated. Three different databases (digits, musical symbols and letters dataset) are considered, and small world, correlated models, and the deactivation network topologies were taken into account. We show that the topology has a definite influence on the performance of the network at short and middle learning time scales. The addition of long-range connections in the deactivation model tended to reduce accuracy. In addition, we examined the effect of degree-degree correlations, or assortativity, on SOM neural networks. It has been shown that high assortativity values tend to decrease the classification error.

Citation

@article{Pimenta2018The,
  title   = {The impact of Interconnecting Topologies on SOM Neural Networks},
  author  = {Pimenta, Mayra Z. and Comin, Cesar Henrique and Rodrigues, Francisco A. and Costa, Luciano Da F.},
  journal = {2018 International Joint Conference on Neural Networks (IJCNN)},
  year    = {2018},
  volume  = {1},
  number  = {},
  pages   = {1--6},
  doi     = {10.1109/ijcnn.2018.8489044}
}