In this lesson
- It starts with a question. The skipper’s neuron gives a good tide 3 points and good weather 2 points, and he goes out if the total passes 4. One morning the weather is good but the tide is not, so the total is 2 and he stays in. It turns out to be a great fishing day. Which change would have sent him out?
- The idea you will find. Training nudges the weights in the direction that shrinks the error. — You compared the neuron’s answer with what really happened, saw the gap, and moved the one weight that could close it. Training a network repeats this in tiny steps over thousands of examples. Nobody tells it the right weights; each mistake pushes them a little closer.
- How you learn it. By doing: a slider, putting things in order.
- Length. 6 short steps. Progress saves on this phone.
Part of Neurons & Networks
Inside the model: how a few numbers added up become a yes or a no.