Comprehensive Guide on tanh

Machine Learning

chevron_right

Neural Networks

chevron_right

Activation functions

schedule Jul 1, 2022

Last updated

local_offer

Tags

tocTable of Contents

expand_more

Derivative Code

map

Check out the interactive map of data science

The $\mathrm{tanh}$ curve looks similar to the sigmoid curve in that the shape is s-like. $\mathrm{tanh}$ takes in any real value as input, and outputs a value from $-1$ to $1$.

The following is a graph of $\mathrm{tanh}$ and sigmoid function:

Mathematically, we can represent $\mathrm{tanh}(x)$ like so:

$$\mathrm{tanh}(x)=\frac{e^x-e^{-x}}{e^x+e^{-x}}$$

Derivative

What makes $\mathrm{tanh}$ appealing as an activation function is that the derivative of $\mathrm{tanh}$ is clean:

$$\frac{d}{dx}\mathrm{tanh}(x)=1-\mathrm{tanh}(x)^2$$

The graph of the derivative of $\mathrm{tanh}(x)$ looks like the following:

Code

The code used to generate the plot is as follows:


        
        
            
                
                
                    import numpy as np
import matplotlib.pyplot as plt
plt.style.use('seaborn-whitegrid')
plt.style.use('seaborn')

def sigmoid(x):
    return 1 / (1 + np.exp(-x))

x_tanh = np.arange(-5, 5, 0.1)
y_tanh = np.tanh(x)
plt.figure(figsize=(5,4))
plt.xlabel("x")
plt.ylabel("y")
plt.axhline(0, color="blue", alpha=0.2)
plt.axvline(0, color="blue", alpha=0.2)
x_sigmoid = np.arange(-5, 5, 0.1)
y_sigmoid = sigmoid(x)
plt.plot(x_sigmoid, y_sigmoid, color="red", label="Sigmoid")
plt.plot(x_tanh, y_tanh, color="blue", label="tanh")
plt.legend()

Published by Isshin Inada

Edited by 0 others

Did you find this page useful?

thumb_up

thumb_down

Comment

Citation

Share on Twitter

Share on Facebook

Ask a question or leave a feedback...

thumb_up

thumb_down

chat_bubble_outline

settings

Enjoy our search

Hit / to insta-search docs and recipes!