This Homework contains Basic implementation of Neural Network on MNIST dataset. The main goal is to understand how a neural network works and how to model a neural network
-
utils.py
-- It contains the utility functions. It has Function namedtrain()
which will calculate loss function and backpropagation method. and functiontest()
will test_loss, correct prediction gives the accuracy. finally functionplot()
will calculate theTraining Loss
,Training accuracy
,Test Loss
,Test Accuarcy
from functionstrain()
andtest()
. -
model.py
-- having ClassNet
is basically a model/convolve of neural network architecture with convolutions and fully connected layers and configurations of these layers. The model takes input, applies convolutional operations, ReLU activations, max-pooling, and fully connected layers to output predictions. -
S5.ipynb
-- It contains and usesutils.py
andmodel.py
functions. It also contains downloading MNIST dataset and loading it to my env. Then I started training the MNIST with my Neural Network Architecture. After that I evaluate the performance and plotted a graph for train, test -- loss and accuracies.