A neural network is a computer model made of connected units, called neurons, arranged in layers. It learns to spot patterns in data such as images, text or sound.
How it is built
An input layer takes data, hidden layers transform it and an output layer gives a result such as a label or a prediction.
How it learns
During training, the network compares its output with the right answer and adjusts connection weights to reduce errors.
Quick reference
| Part | Role | Example |
|---|---|---|
| Input | Receives data | Pixels |
| Hidden layer | Finds patterns | Edges |
| Output | Gives result | Cat or dog |
| Weights | Learned numbers | Adjusted in training |
Where you see them
They power photo recognition, voice assistants, translation and language models.
Limits
They need lots of data, can be hard to explain and may copy biases from training data.
Frequently asked questions
Is a neural network a brain?
No, it is only loosely inspired.
What is deep learning?
Neural networks with many layers.
Do they need a lot of data?
Usually.
Can they be wrong?
Yes.