A modular Java neural network implementation.
Using objects to represent a neural network, meant to be easy to understand (for me).
Custom learning algorithms can be implemented for various techniques with the data such as curriculum learning or stochastic gradient descent, with a standard gradient descent being included at the moment.
Written mostly to help me understand the math behind training a neural network as well as practice git and library design with objects and inheritance.
3blue1brown’s deep learning series on youtube
This video on backpropagation
Maven / add library to central repository when v1 is complete
Test implementing a Trainer outside of the packages
Generate JavaDocs