Classifying with a two neuron CNN

We analyze the two neuron CNN for the particular parameter range where the system converges to constant outputs. The functional relation between the external inputs and the steady state values of the neuron states is found and proves to be useful to solve classification problems. In fact, an exhaustive classification of the binary inputoutput relations that can be achieved by a two neuron CNN is established. From this relation, we propose an algorithm relating the CNN parameters and each one of the different classification problems. As an illustration, we attempt to implement the header action of a universal Turing machine and Boolean functions. Our results are compared to the CNN universal cell.