Deep-Learning-using-MNIST-and-CIFAR10-datasets

This notebook consists of image classification and prediction of MNIST and CIFAR10 data sets using Fully connected Neural Networks and Convolution Neural Networks. MNIST consists of 70000 grayscale images where as CIFAR10 consists of 60000 RGB images. I built several training models initially to test accuracy on training set. Later I selected the best model, created a checkpoint, loaded the pre-trained model back in and tested for validation accuracy. Later, I compared performances of CNN and Fully connected neural networks on the CIFAR10 data set. Please have a look and let me know how I can improve my code. Cheers.