AI & Machine Learning · reviewed in August 2026
Neural network
A neural network is a machine learning model made of layers of connected units ('neurons'), where each connection has a weight adjusted during training. Stacking many simple layers lets it approximate very complex functions, from classifying images to generating text.
Frequently asked questions
What does a single neuron do?
It sums its inputs weighted by its connections' weights, adds a bias, and passes the result through a non-linear activation function that decides how 'activated' that neuron ends up.
Why are non-linear activation functions needed?
Without them, stacking layers would mathematically collapse into a single linear transformation, no matter how many layers there were — non-linearity is what lets the network learn complex patterns.
How does a neural network learn its weights?
With backpropagation: it computes the prediction's error, propagates that error backward through the network to figure out how much each weight contributed, and adjusts the weights with gradient descent.