Skip to main content
All terms
Foundations

Sigmoid

An activation function that squashes any input into the range between 0 and 1.

Definition

The sigmoid function maps any real number into the range between 0 and 1 using 1 / (1 + exp(-x)). It is still used for binary classification outputs, where the result reads as a probability. Its drawback in deep networks is the vanishing-gradient problem: for very large or small inputs the gradient is near zero, slowing learning, so ReLU and its variants have largely replaced it in hidden layers.