Passing Messages Along
Have you ever played the telephone game? You whisper something to a friend, they whisper it to the next person, and the message travels down the line until it reaches the end. Sometimes it changes a little along the way! Inside a neural network, something similar happens — but the messages are numbers, and the network is very careful not to mix them up.
How Signals Travel
When a neural network gets information — say, a picture of a cat — it turns that picture into lots of numbers. Each number describes one tiny piece of the picture, like how bright a certain spot is. Those numbers are the starting signals. They travel into the first set of artificial neurons. Each neuron looks at the numbers it receives, does a tiny calculation, and sends a new number forward. That new number is the neuron's signal to the next neuron. Every neuron in the network does this same step: receive a signal, do a tiny calculation, send a result forward. The signals keep traveling until they reach the very last neurons — and those last neurons give us the answer.
A signal is just a number that travels from one neuron to the next. The network passes millions of these numbers around in less than a second to find its answer.
Here is a story to make it concrete. Imagine a relay race at school. The first runner carries a baton and hands it to the next runner, who hands it to the next, and finally the last runner crosses the finish line. Each runner only carries the baton for their own short stretch. In a neural network, the signal is like the baton. Each neuron carries it for one small step, adjusts it a tiny bit, and passes it on. By the time the signal crosses the finish line — the last layer of neurons — the network has used all those tiny adjustments to figure out an answer.
Fill in the blank to complete the sentence about how signals travel.
One important thing: the signal always moves in one direction — forward. From the first neurons, through the middle, and out to the last neurons. It is like a river flowing downstream. The water never flows back uphill on its own. Scientists call this forward direction the forward pass, because the signal passes forward through the network.
Signals in a neural network always move forward — from start to finish. This forward journey is how the network turns raw information into an answer.
What is a signal inside a neural network?
In which direction do signals travel through a neural network?
The Number Relay
- Sit in a line with two or more people.
- The first person thinks of a number between 1 and 10. That is the starting signal.
- They whisper it to the next person, who adds 1 to it, then whispers the new number to the next person.
- The last person says the number out loud.
- Together, figure out how many people changed the number along the way.
- Now try it with subtracting 1 each time, or doubling the number.
- Notice how each person only did ONE simple math step, but the number changed as it traveled down the line. That is exactly how signals move through a neural network!