Project Description The project aims to boost digit classification accuracy on the MNIST dataset using convolutional neural networks (CNNs). Leveraging my expertise, I've implemented the LENET CNN architecture with batch normalization and dropout layers to mitigate overfitting. The code loads and preprocesses the dataset, constructs the CNN model, trains it for 20 epochs, and evaluates accuracy. My contributions also include visualizing training progress and testing the model with a drawn image for prediction. While the current implementation is robust, potential enhancements could involve exploring different CNN architectures, tuning hyperparameters, and optimizing techniques to further improve accuracy.

#contributors Suraj Upadhayay