this program is about neural network that uses backpropagation method for weights update.

dataset use : mnsit handwritten digits dataset. link:http://yann.lecun.com/exdb/mnist/

how to run: place the dataset (4 files) in same directory where you are running the .py file.

backporpagation: in this method the weight update is done from output layer to hidden and then hidden to input layer. the updation is carried out in a batch (batch update).

classes: there are 10 classes each for single digit from 0 to 9.

in this program i am also compairing the result of neural network with k nearest negihbours and also validating the neural network with the 5 fold cross validation rule.