← Machine Learning & Deep Learning

⏱ About 10 min✦ 10 XP

You Practice, Machines Practice

You and a learning machine have more in common than you might think. You both get better at things by practicing. You both make mistakes along the way. You both need good feedback to improve. But there are also some big differences! Today we will look at both sides — what is the same and what is different about how you learn versus how a machine trains.

What You Have in Common

When you practice the piano, here is what happens: You play a note. Maybe it sounds wrong. You notice the mistake. You adjust your finger placement. You try again. Slowly, your fingers learn where to go without your brain having to work as hard. When a machine trains on examples, here is what happens: It looks at one example. It makes a guess. The guess is wrong. It notices the error. It makes a tiny adjustment inside itself. It looks at the next example. Slowly, it gets better at making correct guesses. Notice anything? Both of you are running the same loop: try, notice the mistake, adjust, try again. The loop is the same. The doer is different.

The Big Idea

Whether it is a child or a machine, getting better follows the same loop: try, notice mistakes, adjust, try again. Practice is practice — the details are different, but the idea is the same.

Let us put the similarities and differences side by side. Both you and a machine start out bad at new things. Both of you need many repetitions to improve. Both of you do better when the feedback you get is accurate. Both of you can get worse if given wrong or confusing feedback. But here are the differences. You feel things — the satisfaction of getting it right, the frustration of making mistakes. A machine feels nothing at all. You can practice slowly, think it over, sleep on it, and come back tomorrow. A machine can run through millions of examples in seconds but does not understand what it is doing. You can ask WHY something is wrong and understand the explanation. A machine just adjusts numbers — it does not understand why.

Match each word to what it means.

Terms

Practice
Training
Mistake
Feedback
Example

Definitions

Doing something over and over to get better at it
Information that tells you if you were right or wrong
When a machine practices on examples to get better
A wrong try that you can learn from
One thing a machine studies while it is learning

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

There is one more big difference worth knowing about. When you learn to ride a bike, you understand what a bike is. You know it takes you places. You know it is fun. You chose to learn. When a machine learns to recognize bikes in photos, it does not know what a bike is. It does not know what riding feels like. It just knows patterns in numbers that correspond to bike shapes. It has no idea why it is learning or what it means. You learn with understanding. A machine learns with pattern matching. Both are powerful — but they are very different kinds of knowing.

You Are Still Special!

Machines can practice millions of times faster than you. But you understand what you learn. You can use it in new creative ways no one ever showed you. You can explain it, change it, and build on it. That is something machines are still not very good at.

What does both a child and a learning machine do when they make a mistake?

Which of these is something ONLY a human learner experiences?

Side-by-Side Practice Journal

  1. Pick a skill you are practicing right now — reading faster, shooting hoops, drawing, anything.
  2. For one week, keep a short journal. Each day write:
  3. Day number
  4. What you tried
  5. One mistake you noticed
  6. One adjustment you made
  7. At the end of the week, look at your journal. Can you see the guess-check-adjust loop in your own practice?
  8. Talk about it: in what ways does your practice journal look like a machine's training log? In what ways is it totally different?
← Back to Machine Learning & Deep Learning