The Sigmoid Activation Function - Python Implementation | DigitalOcean

While we believe that this content benefits our community, we have not yet thoroughly reviewed it. If you have any suggestions for improvements, please let us know by clicking the “report an issue“ button at the bottom of the tutorial.

In this tutorial, we will learn about the sigmoid activation function. The sigmoid function always returns an output between 0 and 1.

After this tutorial you will know:

What is an activation function?
How to implement the sigmoid function in python?
How to plot the sigmoid function in python?
Where do we use the sigmoid function?
What are the problems caused by the sigmoid activation function?
Better alternatives to the sigmoid activation.

Mục Lục

What is an activation function?

An activation function is a mathematical function that controls the output of a neural network. Activation functions help in determining whether a neuron is to be fired or not.

Some of the popular activation functions are :

Binary Step
Linear
Sigmoid
Tanh
ReLU
Leaky ReLU
Softmax

Activation is responsible for adding non-linearity to the output of a neural network model. Without an activation function, a neural network is simply a linear regression.

The mathematical equation for calculating the output of a neural network is:

Sigmoid Activation Function formula Activation Function

In this tutorial, we will focus on the sigmoid activation function. This function comes from the sigmoid function in maths.

Let’s start by discussing the formula for the function.

The formula for the sigmoid activation function

Mathematically you can represent the sigmoid activation function as:

Formula Formula

You can see that the denominator will always be greater than 1, therefore the output will always be between 0 and 1.

Implementing the Sigmoid Activation Function in Python

In this section, we will learn how to implement the sigmoid activation function in Python.

We can define the function in python as:

import
 numpy as np 
def
 sig
(
x)
:
 return
 1
/
(
1
 +
 np.
exp(
-
x)
)

Let’s try running the function on some inputs.

import
 numpy as np 
def
 sig
(
x)
:
 return
 1
/
(
1
 +
 np.
exp(
-
x)
)


x =
 1.0
print
(
'Applying Sigmoid Activation on (%.1f) gives %.1f'
 %
 (
x,
 sig(
x)
)
)

x =
 -
10.0
print
(
'Applying Sigmoid Activation on (%.1f) gives %.1f'
 %
 (
x,
 sig(
x)
)
)

x =
 0.0
print
(
'Applying Sigmoid Activation on (%.1f) gives %.1f'
 %
 (
x,
 sig(
x)
)
)

x =
 15.0
print
(
'Applying Sigmoid Activation on (%.1f) gives %.1f'
 %
 (
x,
 sig(
x)
)
)

x =
 -
2.0
print
(
'Applying Sigmoid Activation on (%.1f) gives %.1f'
 %
 (
x,
 sig(
x)
)
)

Output :

Applying Sigmoid Activation on (1.0) gives 0.7
Applying Sigmoid Activation on (-10.0) gives 0.0
Applying Sigmoid Activation on (0.0) gives 0.5
Applying Sigmoid Activation on (15.0) gives 1.0
Applying Sigmoid Activation on (-2.0) gives 0.1

Plotting Sigmoid Activation using Python

To plot sigmoid activation we’ll use the Numpy library:

import
 numpy as np
import
 matplotlib.
pyplot as plt
x =
 np.
linspace(
-
10
,
 10
,
 50
)   
p =
 sig(
x)
plt.
xlabel(
"x"
) 
plt.
ylabel(
"Sigmoid(x)"
)  
plt.
plot(
x,
 p) 
plt.
show(
)

Output :

Sigmoid Sigmoid

We can see that the output is between 0 and 1.

The sigmoid function is commonly used for predicting probabilities since the probability is always between 0 and 1.

One of the disadvantages of the sigmoid function is that towards the end regions the Y values respond very less to the change in X values.

This results in a problem known as the vanishing gradient problem.

Vanishing gradient slows down the learning process and hence is undesirable.

Let’s discuss some alternatives that overcome this problem.

ReLu activation function

A better alternative that solves this problem of vanishing gradient is the ReLu activation function.

The ReLu activation function returns 0 if the input is negative otherwise return the input as it is.

Mathematically it is represented as:

Relu Relu

You can implement it in Python as follows:

def
 relu
(
x)
:
    return
 max
(
0.0
,
 x)

Let’s see how it works on some inputs.

def
 relu
(
x)
:
    return
 max
(
0.0
,
 x)
 
x =
 1.0
print
(
'Applying Relu on (%.1f) gives %.1f'
 %
 (
x,
 relu(
x)
)
)
x =
 -
10.0
print
(
'Applying Relu on (%.1f) gives %.1f'
 %
 (
x,
 relu(
x)
)
)
x =
 0.0
print
(
'Applying Relu on (%.1f) gives %.1f'
 %
 (
x,
 relu(
x)
)
)
x =
 15.0
print
(
'Applying Relu on (%.1f) gives %.1f'
 %
 (
x,
 relu(
x)
)
)
x =
 -
20.0
print
(
'Applying Relu on (%.1f) gives %.1f'
 %
 (
x,
 relu(
x)
)
)

Output:

Applying Relu on (1.0) gives 1.0
Applying Relu on (-10.0) gives 0.0
Applying Relu on (0.0) gives 0.0
Applying Relu on (15.0) gives 15.0
Applying Relu on (-20.0) gives 0.0

The problem with ReLu is that the gradient for negative inputs comes out to be zero.

This again leads to the problem of vanishing gradient (zero-gradient) for negative inputs.

To solve this problem we have another alternative known as the Leaky ReLu activation function.

Leaky ReLu activation function

The leaky ReLu addresses the problem of zero gradients for negative value, by giving an extremely small linear component of x to negative inputs.

Mathematically we can define it as:

f(
x)
=
 0
.
01x,
 x<
0
    =
 x,
   x>=
0

You can implement it in Python using:

def
 leaky_relu
(
x)
:
  if
 x>
0
 :
    return x
  else
 :
    return
 0.01
*x
  
x =
 1.0
print
(
'Applying Leaky Relu on (%.1f) gives %.1f'
 %
 (
x,
 leaky_relu(
x)
)
)

x =
 -
10.0
print
(
'Applying Leaky Relu on (%.1f) gives %.1f'
 %
 (
x,
 leaky_relu(
x)
)
)

x =
 0.0
print
(
'Applying Leaky Relu on (%.1f) gives %.1f'
 %
 (
x,
 leaky_relu(
x)
)
)

x =
 15.0
print
(
'Applying Leaky Relu on (%.1f) gives %.1f'
 %
 (
x,
 leaky_relu(
x)
)
)

x =
 -
20.0
print
(
'Applying Leaky Relu on (%.1f) gives %.1f'
 %
 (
x,
 leaky_relu(
x)
)
)

Output :

Applying Leaky Relu on (1.0) gives 1.0
Applying Leaky Relu on (-10.0) gives -0.1
Applying Leaky Relu on (0.0) gives 0.0
Applying Leaky Relu on (15.0) gives 15.0
Applying Leaky Relu on (-20.0) gives -0.2