On cellular neural network learning
X. Vilasis-Cardona, Mireia Vinyoles-Serra · 2006
We study the continuous time two neuron CNN in the context of a supervised learning problem and extract the convergence map for constant inputs. From this result we can design a CNN cloning template for a classification problem. Some of the parameters of this cloning template play no effective role in the tesselation giving the convergence map. This gives some hints on how to set up a learning problem and solves the template composition problem for this case.