Phase 4 · Deep learning · Neural networks
Multi-layer networks
Stack layers to learn complex functions.
Open this lesson Create a free account
Why it matters
One neuron draws one line. A layer of neurons draws many lines at once, and the next layer combines them into curves, corners and islands. That's the multi-layer perceptron (MLP), the basic neural network, and it isn't only historical: every transformer block in an LLM contains one, and those MLP blocks hold most of the model's parameters.
What you'll be able to do
- Describe a layer as many neurons computed at once with a matrix multiplication
- Solve XOR with a two-layer network and explain how the hidden layer does it
- Count a network's parameters from its layer sizes
- Explain what depth and width add, informally (universal approximation)
Before this lesson
Next in this phase: The forward pass