Use of the CNN dynamic to associate two points with different quantization grains in the state space

The paper is concerned with the design of a part of the CNN state space trajectory. A point in the CNN state space represents a sampled signal (the state of each neuron is a sample): the set of points generated by the CNN state evolution can thus represent a set of sampled signals. We describe a methodology which allows us to find the initial state and the CNN weights so that the CNN state evolution is, at a fixed time t₀, as close as possible to the point representing a given sampled signal. In such way a signal is described through the CNN initial state, the cloning template and the time instant t₀. In order to find the CNN initial state and the CNN weights we used a procedure based on Genetic Algorithms