Character recognition is the main part of this
system for the identification of character. Here the
traditional optical character recognition is replaced by
feed forward, Back Propagation neural network for a
given system. To adjust the connection weights BPLAs
use the gradient-decent search method. In feed-forward
neural network information moves only in one forward
direction. The direction is from the input layer via the
hidden layer, to the output layer. Loops are not present
in this network. The limitation of single layer
perceptrons has been overcome by applying multilayer feed-forward network with one or more hidden layers.
It can be trained using back propagation algorithms.
This increases the overall recognition rate due to multi layer feed forward network. Randomly, weights are chosen, the needful corrections can be computed using back propagation algorithm. This algorithm doesn’t works when the value of error function becomes negligible small.