Dynamic Self-Organized maps

This is a variation of the self-organising map algorithm where the original time-dependent (learning rate and neighbourhood) learning function is replaced by a time-invariant one. This allows for on-line and continuous learning on both static and dynamic data distributions. One of the property of the newly proposed algorithm is that it does not fit the magnification law and the achieved vector density is not directly proportional to the density of the distribution as found in most vector quantisation algorithms. From a biological point of view, this algorithm sheds light on cortical plasticity seen as a dynamic and tight coupling between the environment and the model.

Reference

Dynamic Self-Organising map N.P.Rougier and Y.Boniface, Neurocomputing 74, 11, (2011), 1840-1847.