MINIST Image Digit Recognition by means of a MPL fully connected neural network
This code creates and trains an MLP fully connected network (three layers) to make predictions about a handwriten digit image type from MNIST dataset. Then, it predicts the probability of an input image to be any from 0 to 9 digits.
More about MNIST dataset (National Institute of Standards and Technology) here: https://en.wikipedia.org/wiki/MNIST_database
This code is an evolution of the code originally developed in the framework of the Module 2, Neural Networks, Lesson 7, Deep learning with PyTorch, of the Udacity NanoDegree program Deep Learning
Just clone the repository in your working directory, open the Jupyter Notebook and run it.