Course 4, lesson 34 of 100, Ages 10+
Neural networks
Tiny deciders that vote
Like I’m 5
Imagine a big team passing notes. Each person reads a note, adds a little, and passes it on. At the end, the team agrees on an answer.
The big idea
A neural network is made of many tiny units in layers. Each unit takes in numbers, multiplies them by weights, adds them up and passes the result on, a bit like passing notes along a team.
Training adjusts the weights so the whole team gives the right answer more often. One layer might notice edges, the next shapes, the next whole objects. The network is loosely inspired by neurons in the brain, but much simpler.
Examples
- Handwritten numbers: A small network can learn to read digits from 0 to 9.
- Voice assistants: Networks turn sound waves into words.
- Weights: A weight is like how much one teammate trusts another's note.
Check your understanding
- What does each dot in a neural network do?
- Options: Passes a signal on to the next layer; Takes a photo; Stores a song.
Answer: Passes a signal on to the next layer. Each neuron mixes the signals coming in and passes its own signal on. - What changes when a neural network learns?
- Options: Its weights; The number of screens; Its colour.
Answer: Its weights. Training tweaks the weights so outputs get closer to the right answers.
Remember
Neural networks pass signals through layers of tiny deciders.
Talk about it
How does your family or team decide things together?
Go deeper
Each dot (a “neuron”) adds up signals from the layer before, weighted by numbers learned in training. Real networks have millions of neurons and many layers, which is why it’s called deep learning.