Tip: press / anywhere to search. Or open a random note →

Handwritten one-page note: How Machine Learning Learns
Tap the note to see it full size

AI & Tech One-page note

How Machine Learning Learns

How does a computer learn without being told the rules? Here's machine learning, explained with a spam filter. Save this page.

Good to know

Simplified. Sources: Royal Society (2017), Google Machine Learning Crash Course. Checked Nov 2026.

What's on the page

The note as text, so you can read it, copy it or use a screen reader.

The big idea

  • No step-by-step rules written by people
  • Learns directly from examples and data
  • Then makes predictions on new data

Example: a spam filter

  • Feed it lots of emails marked "spam" or "not spam"
  • It looks at subject, text and sender
  • Finds patterns that tend to mean spam

Training

  • Starts with random settings
  • Guesses, measures how wrong it was
  • Nudges its settings to be less wrong
  • Repeats until it stops improving

Testing

  • Test it on emails it has never seen
  • New examples are the only fair test
  • Perfect on training, bad on new? It memorised (overfitting)

You use it every day

  • Fraud alerts on your bank card
  • Photo tagging and translation
  • "You might also like" suggestions

Sources

  1. Royal Society (2017): Machine learning, the power and promise of computers that learn by example. Summary
  2. Google for Developers: What is machine learning?
  3. Google ML Crash Course: Gradient descent
  4. Google ML Crash Course: Dividing the original dataset
  5. Google ML Crash Course: Overfitting

How we checked this

Every fact on this note comes from the 5 sources listed here, checked against official or peer-reviewed sources where possible. AI tools help with drafting; a person checks every fact against the sources. How we research

Spotted a mistake? Tell us and we'll fix it and log it on the corrections page.

More from AI & Tech

All AI & Tech notes →