A fun demonstration of simple neural networks
The only neural network in this file is a simple 3 layer network to give the correct outputs to the XOR boolean operator (0,0 -> 0) (0,1 -> 1) (1,0 -> 1) (1,1 -> 0)
Since there is no linear combination that gives these outputs, it is an ideal example case for a concept demonstration of a neural network.
You will notice in the file that the partial derivatives for the back-propagation are worked out explicitly and then hardcoded instead of doing the calculus numerically. This is not very conventional, however, understanding the mathematics behind back-propagation can make us more versatile and adaptable when working with neural networks and machine learning.
(The others that I have worked on are not fully functional but will be added in the near future, namely 2 sets of 2 neural networks that play tic-tac-toe and Connect 4 against each other)