Two-dimensional memristive CNN for sequence recognition
Zbigniew M. Wawrzyniak
The sequence pattern recognition can be performed by cellular neural network with delayed templates. The CNN with memristive cells can be applied directly without delays by direct exploration of memristor dynamics. The paper demonstrates the general idea by using the Bernoulli model of memristor. We demonstrate the examples of pixel movements that can be recognized by the trace of its past positions. Some general conditions must be satisfied as the pixel speed with respect to memristor dynamics.