Build and train your own neural network to detect your hand-written digits in the digit-detection app.
Digit Classification is an app where you can fully customize and build your neural network using the network model code and parameter database provided to experiment and study how neural networks learn. I started working on this project to learn how machines could possibly learn and how neural networks function to help machines learn. Essentially, a neural network is just an algorithm that uses mathematics to teach the machine what results are wanted and what results are unwanted. By using a gradient descent algorithm we can train the neural network to minimize it's penalties (in other words to decrease it's cost) and manipulate it's settings. During this whole project, I found neural networks very fascinating and sophisticated and I can't believe how easy it is for a machine to learn with just some minimal multi-variable calculus.
DISCLAIMER: This was my first ever implementation of a Neural Network from scratch so the code is really messed up and nasty. I wouldn't use the code as a resource to learn how to code a Neural Network but the application itself works perfectly fine.
Train and Reset Neural Network in the application to play around with the Machine Learning algorithm.
