Training Is Like Practice

You've seen your best friend's notes so many times that you'd recognise their handwriting anywhere — even on a page you've never read.

Nobody gave you a rulebook. You practised by seeing lots of examples. That's exactly how a machine learns.

Almost all machine learning follows the same four steps:

examples → pattern → model → prediction

• Examples — the data you show it
• Pattern — what those examples have in common
• Model — the 'learned recogniser' it builds
• Prediction — its best guess on something new