This Repo has deep learning classifiers to recognize MNIST Digits from 0 to 9.
It is a notebook that plots MNIST dataset that has 784 pixels and 42000 datapoints in 2 Dimensions using PCA. I have used PCA to squash 784 features into 330 features that retains 99% of the variance.
It is a CNN classifier that predicts the digits with 99% accuracy.
Training Loss vs Validation Loss
CNN Architecture
My Kaggle Notebook - https://www.kaggle.com/shweta2407/mnist-digit-recognition-using-cnn-0-99-accuracy