The human brain is extraordinarily impressive, it has been studied for centuries, but is so incredibly complex that even now it is still barely understood. The human brain contains around 86 billion neurons, each neuron is connected to thousands of other neurons through connections called synapses which allow for signals to be passed through the brain.

This project was created to build a digit recognition neural network, it can classify the 10,000 MNIST test images with an accuracy of 96%, a GUI has also been implemented which allows us to create new networks with adjustable parameters, test a network with a mouse drawn image on a ”paint” like pad, an image selected from a file and also the MNIST test database.

Demo: