• RepositoryGitHub
• LanguagePython [Pygame], Jupyter Notebook

In this project, I endeavored to create Hopfield neural network. These are a type of neural network designed to be capable of storing and retrieving many multitudes of memories. I created an implementation of a Hopfield network from scratch in a Python Jupiter Notebook. The network was trained on the MNIST handwritten dataset and used PyGame to visualize the output. I created two ways to test the capabilities of the network, firstly you can simply ask the network to draw a number. This will have the access all its memories regarding that number, and create a unique drawing of it. Secondly, you can input a scrambled up version of a number. This will have the network use it's memories of numbers to create a best guess for the original unscrambeled number.