Your brain already does this
You are leaving for school. Big decision: take the umbrella, or leave it?
You check the clues. Dark clouds? Rain yesterday? Did Grandma say walk with it? Some clues matter a lot. Some matter a little. Dark clouds matter a LOT. And Grandma? Grandma matters a lot too. 😄
A robot brain decides the same way. It gives each clue importance points. Then it adds them up. Big total? YES! Small total? NO. That little decider is called a neuron.
A tiny decider inside a robot brain. It takes clues, multiplies each one by its importance points, adds everything up, and compares the total to a magic number to decide yes or no.
The robot's maths (you can do this)
Here is the whole trick, with real numbers. Each clue is worth 1 if it is true and 0 if it is false. Then you multiply by the importance points and add:
Try it! Dark clouds = 4 points. Rain yesterday = 2 points. Grandma said nothing = 0 points.
Add them: 4 + 2 + 0 = 6. Is 6 at least 5? YES! So the neuron shouts: take the umbrella! The "5" is the magic number. Its fancy name is the threshold.
It is like deciding if the pot of soup is ready. Smells done (3 points), been cooking one hour (2 points), dumplings float (4 points). Add up the points from the clues that are true, and if the total beats the magic number, lunch time! Your nose and eyes have been doing neuron maths your whole life.
🎮 Mini game: YOU are the neuron
Tap a clue's OFF button to switch it ON, and watch the total change. The umbrella answer updates as you go. Can any single clue reach 5 all by itself?
The clues are: dark clouds ON (4 points) and radio showers ON (1 point). The magic number is 5. What does the neuron decide?
Tap the answer you think is right.
From one neuron to a whole brain
One neuron makes one small decision. Now the cool part: neurons can pass their answers to OTHER neurons! Stack them in layers, like a big team, and you get a neural network. "Deep" learning just means the team has many layers.
One neuron asks "dark clouds?". A team of neurons asks "rainy day?". A BIG team can tell a crab from a fish, or guess tomorrow's weather. Same tiny maths, done super fast.
A team of neurons working in layers. Each layer passes its answers to the next, so simple little decisions grow into big clever ones. "Deep learning" means the team has many layers.
But who picks the importance points?
Great question! A brand-new robot brain just guesses. It makes silly mistakes. Then it learns: guess, check how wrong, fix the points. Again and again, millions of times. That is called training. In the next two games, YOU do the guessing, and your score shows how close you got!
1) A neuron multiplies clues by importance points, adds them, and compares with a magic number. 2) A neural network is layers of neurons passing answers forward. 3) The network learns by guessing, checking how wrong it was, and fixing its points.
🎮 3D Vox game: Pulse!
Now SEE the neuron work! Switch clues ON and watch glowing pulses race along the wires into the brain. Fill the charge to 5 and it FIRES: the lamp lights up!
Then it is game time: go be a classifier! 🦀