In this lesson
- It starts with a question. A fishing decision uses two signals: tide (worth 3) and weather (worth 2). Both are good, and the skipper goes out if the total passes 4. Does he go?
- The idea you will find. A weight says how much a signal matters. — Raise the tide weight and tide decisions dominate; drop it to zero and tide is ignored. Training a network is nothing more than nudging thousands of these weights until the decisions come out right on the examples. There is no reasoning inside — only weighted sums.
- How you learn it. By doing: a slider.
- Length. 5 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.