Neural_network_tracking

Nowadays neural networks are widely used in many branches of physics, in particular in particle physics.
The Higgs ML challenge asked to classify in signal and background a set of events simulated based on the ATLAS detector. This is exactly the kind of problem that could be effectively tackled using a neural network.
Nowadays there are a lot libraries already written to work with neural networks. The first two that come to mind, and also the largest ones, are Tensorflow and PyTorch, which are available in Python, C++ and other languages.
These libraries are very thoroughly written and efficient, but from an accademic point of view it is instructive, in order to really understand the functioning of neural networks, to learn how to write them from scratch.
The goal of this project is to write from scratch in C++ a framework for building neural networks, test it with the MNIST dataset and finally use it to tackle the Higgs ML challenge.