This project was aimed for the classification of Handwritten digits using the MNIST Dataset. The project was implemented in two phases .
-
Using Numpy and Writing code for Forward Propagation and Backward Propagation from scratch.
-
Using Convolutional Neural Networks with Pytorch Framework.
WIth Numpy accuracy obtained are
-Training accuracy of 96.73 %
-Test accuracy of 96.21 %
With CNN accuracy obtained are
-Training accuracy of 99%
-Test accuracy of 98.98%