This is a C++ implementation of a basic, general-purpose feedworward neural network with a freely configurable number of layers and freely configurable number of neurons in each layer.

The main program's purpose is to train a NN to detect handwritten digits as provided by the MNIST dataset . Since the MNIST dataset contains 28x28 pixel large annotated pictures of handwritten digits, the input layer of the program's neural network has 28x28=784 neurons. And since there are 10 different digits to be detected, the output layer has 10 neurons.

Github repository: https://github.com/chn-dev/NeuralNetwork