After watching this presentation, where Martin Görner shows the use of recurrent neural networks to create a new Shakespeare play, I was feeling inspired. So, I pointed his code a bunch of song lyrics.

Background

I’ve been interested in Neural Networks for a long time because, fundamentally, they are similar to a process controller.

$$ Output = Input \times Weight + Bias $$

It’s just a line. For multiple inputs/outputs, the equation is the same, but they are matrices (or tensors). Why stop there, send the output of one “neuron” to the input of another, and you’ve got a “deep” network.