Classifiers for decoding patterns in the response of an artificial retina

We present a neural classifier able to reliably determine the correlation between elementary image patterns and the characteristics of the chaotic regime of an artificial retina based on nonlinear dynamics. While the classifier is a typical neural one, it has the role of completing the neural retina with the equivalent of the upper layers of neurons in the nervous system of mammals. Without this classification stage, the proper use of the retina is imperfect, hence the significance of reporting on it. The performance of the classifier is discussed in relation to the application requirements. This paper represents a preliminary, abstracted version of a paper to be published in a journal.