A single-layer perceptron is the basic unit of a neural network. A perceptron consists of input values, weights and a bias, a weighted sum and activation function. And with such a simple construction, a perceptron can "learn" a complicated linear function and solve it.

Inspired by Deep Mind's astonishing feats of having their Alpha Go, Alpha Zero and Alpha Star programs learn (and be amazing at it) Go, Chess, Atari games and lately Starcraft; I set myself to the task of programming a neural network that will learn by itself how to play the ancient game of tic tac toe. How hard could it be?