Digits and Special Character Recognition System using ANN and SNN Models

The paper depicts pattern recognition of Digits and Special Character using Artificial Neural Network’s model Feed Forward Neural Network using Back Propagation Algorithm and Spiking Neural Network’s model Izhikevich Neuron model. Artificial Neural network is the second generation model and Spiking Neural Network is third generation model of Neural Networks. In this paper we have focused on recognizing the patterns and comparing the results of the two models Feed forward Neural Network and Izhikevich model by following the main steps of pattern recognition which are, Scanned handwritten Images, Pre-Processing of image, Feature Extraction of image, Training the network, Recognition/Classification of images.

Feed Forward Neural Network consists of three layers and this network is trained by Back propagation Algorithm. On the other hand, we have Izhikevich model which is well known for producing all known firing rates pattern. It is a combination of Hodgkin-Huxley model and computationally efficient Integrate-and Fire neuron model.  Izhikevich model will recognize the input pattern of the image and generate the spikes. Using the above two models simulation results are obtained for recognition of digits and special characters and then comparing both models based on recognition rate, reliability, simulation time and number of neurons. We show that the Izhikevich model of spiking neural type provides improved performance.