SERVING TWO · THE THREE THINGS MACHINES LEARN · LESSON 2.1 OF 19 · ABOUT THREE MINUTES
The video for this lesson is in production. The full narration is below, and it reads as Ah Liew would speak it.
Watch or read
Five in the morning. The wet market is already loud. Ah Liew stands at the prawn stall with a basket between his feet, and he has a job that cannot wait. Every prawn must be sorted before it touches his stock. Fresh goes in. Spoiled goes out. There is no machine, no label, no time. He picks one up, turns it, smells it, and drops it left or right. Fresh. Spoiled. Fresh. He does this five hundred times before the sun is fully up.
Ah Liew cannot give you the rule. Ask him how he knows, and he will say, “I just know.” What he means is: he has held tens of thousands of prawns, and each one taught his hand and nose a little more. His head now holds a library of labelled examples, this one fresh, that one spoiled. When a new prawn arrives, he never starts from scratch. He asks one question: which of the prawns I already know does this one look like?
That is classification. It is the most useful thing a machine can learn. You give it hundreds or thousands of examples, each one already labelled: fresh or spoiled, fraud or genuine. It builds the same kind of library Ah Liew carries in his head. Then you hand it something new, and it sorts it.
This is not a hawker trick. It is everywhere. Your email does it before you wake: this message is spam, this one is real, and you never even see the ones it catches. Your bank does it the instant you tap your card in another country. Is this really you, or is someone using your number? A farm does it on a belt: this fruit is ripe, this one is not, sorted at speed. Same idea, every time: labelled examples in, a sort on something new out.
TRY THIS · PAUSE AND PREDICT
Let me hand you a prawn. Firm. Glossy shell. Smells faintly of the sea. Fresh, or spoiled? The video gives you three seconds. Take thirty if you want. This page will not move until you do.
Show Ah Liew’s answer
Fresh. You matched what you felt against prawns you have known your whole life. That is all a classification model does: it holds a library of labelled examples, and it asks, which of these does the new one look like?
And tonight, when you sort the laundry, darks from lights, or the good container from the recycled one, you will feel a little of Ah Liew in yourself. Learned labels, new items, sorted.
IN ONE LINE: Classification is a library of labelled examples and one simple question: what does this new thing resemble?