A disclaimer: my understanding of neural networks (along with my understanding of the entire field of data science and, you know, the world) is a work in progress. Despite my best efforts, some of the explanations in this post will lack context, misrepresent nuance, or just be flat out wrong. This post is not meant to be a comprehensive explainer on the topic of neural networks - I am writing this for my own benefit as much as anyone else's, and as an attempt to capture my understanding of the subject at this point in time.
That said, on DAY ELEVEN, we can finally start answering the question WHAT IS A NEURAL NETWORK?
If you could not tell, I AM SO EXCITED. 🎉🎉🎉
According to Dr. Google, a neural network is a computational system modeled on the human brain.
That sounds pretty cool. So how does the human brain serve as inspiration for the design of neural networks? I guess neural networks have layers, and nodes, and brains have... Layers. And nodes. Maybe.
Ok, so despite several years of undergraduate psychology education I don't actually know how the human brain serves as inspiration for the design of neural networks. I've made a note to find out. Instead of worrying about the how, today we're going to focus on the what.
Borrowing heavily from the fast.ai approach, we're going to build an understanding of neural networks by implementing one in a spreadsheet and walking through it one step at a time. I've embedded the spreadsheet below, but I've put the original here so you can mess around with the formulas and stuff.