AlgoPlus//machine learning / mlp
Read the theory

Multilayer Perceptron

Stacked layers of neurons transform the input into an output, one layer at a time.

Layers
Active
Legend
Activation magnitude
Active layer
AI Tutor Workspace
In a nutshell
A multilayer perceptron is a stack of layers of neurons. Each neuron multiplies its inputs by learned weights, adds them up, and passes the total through a non-linear activation. Feeding numbers in at one end and reading the answer off the other is a forward pass. The hidden layers in the middle let the network learn its own features, so it can model curved, complex relationships a single straight line never could.
Ready
Press play to begin the cinematic walkthrough.
Each layer is a fresh set of feature detectors built from the last. Stack enough non-linear layers and the network can bend its decision surface into almost any shape.
Key terms
Go deeper in the lesson
Read the full theory, intuition & complexity for Multilayer Perceptron.