direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

All Publications

Large-Scale Simulation of a Self-Organizing Neural Network: Formation of a Somatopic Map
Citation key Obermayer1990
Author Obermayer, K. and Ritter H., and Schulten, K.
Title of Book Parallel Processing in Neural Systems and Computers
Pages 71 – 74
Year 1990
Editor Eckmiller, R. and Hartmann, G. and Hauske, G.
Publisher Elsevier
Abstract The "somatotopic map" of the body surface of animals and humans reflects an ordered, neighborhood preserving connectivity between tactile skin receptors and cortical neurons. The border between adjacent areas representing different body parts changes even in adult animals, which shows, that the map is dynamically maintained, and makes this system an interesting model for the investigation of cortical plasticity. In this paper we present a large-scale simulation study of a neural network model for the formation and readaptation of a somatotopic map of the inner hand surface. The neural network, which is based on an algorithm developed by Kohonen [3,4], contains 16,384 neurons and 800 tactile receptors, giving a total of 13,107,200 adaptive connections. The simulation assumes random initial connections between the receptors and neurons, and the formation of the map proceeds during randomly applied, local stimuli. The number of necessary adaptive steps was surprisingly small, which indicates, that a high dimensional input space facilitates the ordering process. Neurons with double and multiple receptive fields can emerge during simulation. The model network readapts upon partial deprivation of sensory input much in the same way as is found in experiments [8].
Link to original publication Download Bibtex entry

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions