AlgoPlus//machine learning / activations
Read the theory

Activation Functions

The non-linearity between layers — without it, a deep net is just one linear map.

Function
ReLU
f(x)
Activation
Legend
f(x)
derivative
AI Tutor Workspace
In a nutshell
An activation function is the bend applied to each neuron's output. It matters because stacking plain linear layers just gives one big straight line; the bend is the non-linearity that lets a deep network fit curved patterns. ReLU keeps positive inputs and zeroes the rest — cheap and the modern default. Sigmoid squashes everything into (0, 1), and tanh into (−1, 1); both are smooth but their gradients fade for large inputs, which can slow learning.
Ready
Press play to begin the cinematic walkthrough.
The kink is the point. A straight line through straight lines is still a straight line — the bend in ReLU or tanh is what lets layers add up to something non-linear.
Key terms
Go deeper in the lesson
Read the full theory, intuition & complexity for Activation Functions.