C++ DNN

This is a simple from-scratch implementation of a feed forward deep neural network in C++ in the form of Matrix and NeuralNet classes. It currently supports stochastic and batch gradient descent, relu, tanh and sigmoid activation functions, and an arbitrary number of hidden layers and nodes.

I enjoy building things from scratch when I'm working on my own projects, as I think it takes me to a level of familiarity with concepts or technologies I couldn't have obtained otherwise. I'm looking to add more features like regularization and command line options in the near future.

View this project on Github