Imagine a tiny light that only switches on when the right things happen. That is the heart of this little game, and it is also the heart of every smart computer you have ever heard about. Meet your one brain cell.
#What you do
In the game you set a few dials on a single pretend brain cell. Each dial is a weight. A weight is just how much one input matters. Turn a dial up and that input counts for a lot. Turn it down and it barely counts at all. You keep nudging the dials until the cell lights up for the right inputs and stays dark for the wrong ones. When it behaves the way you want, you have won that round.
#What is really happening
The brain cell takes each input and multiplies it by its weight. Then it adds all those results into one total. If the total climbs past a line, the cell fires and lights up. If it stays under the line, nothing happens. That is the whole trick. Getting better is nothing more than adjusting the weights so the cell fires at the right times. When a computer does this on its own, we call it learning. You are doing the same job by hand.
#Where it shows up
One brain cell sounds small, and it is. But it is the smallest building block of every neural network. Stack many of these cells together, let each one pass its light to the next, and you can spot a cat in a photo or finish your sentence in a text. Every one of those giant systems is built from the same simple idea you are playing with right now. Want to try it yourself? play it and wiggle the dials.
#The short version
- A brain cell weighs each input, adds them up, and fires if the total passes a line.
- Learning means adjusting the weights until it fires at the right times.
- One neuron is the smallest building block of every neural network.