This code is for a school challenge that I partaked in. This challenge is based off the MNIST (Modified National Institute of Standards and Technology). I had to develop a neural network that can be train and used to accurately decipher various handwritten numbers from 1 to 10.
My approach to this problem was to use a convolutional neural network, so first I had to reshape the dataset to make it fit with my convolutional neural network algorithm. After that, I split my dataset into training and testing dataset with test set to 10 percent. Then for my algorithm, I used max pooling. Overall, after training it, I algorithm can accurately predict the numbers with a around 97% accuracy.