← Machine Learning & Deep Learning

⏱ About 10 min✦ 10 XP

Data Is Just Examples

When you were learning to read, someone showed you the letter A over and over. Big A, little a, red A, blue A, fancy A, plain A. After seeing so many examples, your brain just knew — that shape is the letter A! Machines learn the same way. They study a huge pile of examples, and slowly they learn to recognize patterns.

Examples Are a Machine's Teacher

Think about learning to recognize a dog. No one gave you a rulebook that said 'four legs, fur, tail, barks.' You just saw dogs — big ones, small ones, fluffy ones, spotted ones — and your brain figured it out. A machine learning program works the same way. Instead of rules, it gets examples. The examples are its data. If a machine sees ten thousand pictures labeled 'dog,' it starts to notice what dogs look like. If it sees ten thousand pictures labeled 'cat,' it learns what cats look like. The examples are the teachers.

The Big Idea

Data is not just information — to a learning machine, data is its examples. Every example the machine studies teaches it a little bit more.

Here is a story to make this real: Meet Pixel, a learning program. Pixel wants to learn what a sunny day looks like. Pixel's helpers collect one thousand photos. Five hundred are labeled 'sunny day.' Five hundred are labeled 'not sunny.' Pixel studies every single photo. After a while, Pixel notices that sunny-day photos tend to have lots of bright yellow and blue. Cloudy ones are more gray. Pixel did not read a textbook about sunshine. Pixel just studied examples — and that is data doing its job.

Match each learning situation to the examples that would help.

Terms

Learning what apples look like
Learning how to spell 'cat'
Learning what a baby's cry sounds like
Learning what the number 7 looks like
Learning what rain sounds like

Definitions

Many recordings of babies crying
Thousands of sentences where the word cat is written correctly
Many recordings of rain falling on windows and rooftops
Thousands of handwritten sevens from different people
Many photos of apples of different colors and sizes

Drag terms onto their definitions, or click a term then click a definition to match.

Every learning machine needs examples. Without examples, a machine has nothing to study — it stays blank, like a brand-new notebook. The more good examples it studies, the better it can learn.

You Learn From Examples Too

Next time you learn something new — a new word, a new math fact, a new song — notice how many examples helped you. Learning from examples is something humans and machines both do!

Why does a learning machine need data?

If a machine studies thousands of photos of flowers, what is it most likely learning?

Teach a Friend With Examples

  1. Pick something simple to teach: maybe what a triangle is, or what the color orange looks like.
  2. Do NOT explain it with words or rules.
  3. Instead, collect examples: draw five triangles of different sizes, or find five orange objects.
  4. Show only the examples to a friend or family member.
  5. Ask them: 'What do all of these have in common?'
  6. See if they can figure it out from the examples alone — just like a machine does!
← Back to Machine Learning & Deep Learning