April 2018
Beginner to intermediate
282 pages
6h 52m
English
The sigmoid function is defined as:

Here, x is the value of the weighted sum of the input values. We have seen this function in logistic regression. When x is below zero it drops off, and anything above zero it approximates towards one. Unlike a step function, it is a non-linear activation function. It is mostly used in output layers in a neural network, when we work on classification tasks trying to predict probabilities. The following figure illustrates the sigmoid function:

The next activation function we discuss is ...