In this problem, you will be using a dataset representing digital images of handwritten digits of different

customers of a large retailer. Each digit is represented in the library using $16$x$16$ pixel black$/$white images,

where the value of every pixel is $0$ if it is black, or $1$ if it is white. Hence, we can consider every image in our

dataset as a sample with $16$x$16=256$ binary features, while the labels represent the actual number that the

image represents. For each digit, we have $600$ training samples and $500$ testing samples that you can find on

Blackboard in the assignments folder. The training samples are stored in the files called $$training$\_$features.txt$$

and $$training$\_$labels.txt$,$ where each line in the labels file corresponds to the label $($actual digit$)$ for the

features represented in the same line in the features file. Same format is used to store the test data.

Implement a neural network model with back$-$propagation using the algorithms we discussed in class

and the appropriate data structures. Make sure your implementation allows for different network

structures that the user can provide $($either interactively or through a config file$).$ Your code has to be

implemented from scratch, not using any of the existing libraries and Use your implementation to train a model using the training data from the provided dataset and

report the average precision, recall, and accuracy of your model on the test data

Note: Since you dont have access to the files, just write the code for the implementation.