MNIST-classification

This project is made using the MNIST dataset and the models are trained to classify the handwritten digits. We use logistic regression to classify the digits of the MNIST dataset. As it is a realtively simple problem it can be easily solved by logistic regression and hence we get an accuracy of about 90% with it. But when we approach many more complex problem like classifying the objects of the CIFAR10 dataset we can understand the boundaries of these simple models.