Basics of deep learning, implemented using NumPy.
Areas covered are feedforward neural networks, simple backpropagation and stochastic gradient descent.
This project is merely a collection of experiments and self-written elements in the area of deep learning and does not aspire to be especially efficient or general. All of the code developed for these implementations of deep learning applications serves the purpose of understanding the underlying mechanics better and does NOT challenge any existing frameworks or libraries.
Most of the code is an implementation (albeit with modifications) of algorithms and concepts presented in Deep Learning - Goodfellow, Bengio, Courville.