Python – tensorflow.math.l2_normalize()

TensorFlow is open-source Python library designed by Google to develop Machine Learning models and deep learning  neural networks.

l2_normalize() is used to normalize a tensor along axis using L2 norm.

Syntax: tensorflow.math.l2_normalize( x, axis, epsilon, name)

Parameters:

• x: It’s the input tensor.
• axis: It defines the dimension along which tensor will be normalized.
• epsilon: It defines the lower bound value for norm. Default value is 1e-12. It uses sqrt(epsilon) as divisor if norm<sqrt(divisor).
• name: An optional parameter that defines the name for the operation.

Returns:

It returns a tensor of same shape as x.

Example 1:

Python3

# Importing the library
import tensorflow as tf

# Initializing the input tensor
a = tf.constant([7, 8, 13, 11], dtype = tf.float64)

# Printing the input tensor
print('a: ', a)

# Calculating the result
res = tf.math.l2_normalize(a)

# Printing the result
print('Result: ', res)

Output:

a:  tf.Tensor([ 7.  8. 13. 11.], shape=(4, ), dtype=float64)
Result:  tf.Tensor([0.34869484 0.39850839 0.64757613 0.54794903], shape=(4, ), dtype=float64)

Example 2: This example uses 2-D tensor.

Python3

# Importing the library
import tensorflow as tf

# Initializing the input tensor
a = tf.constant([[7, 8], [13, 11]], dtype = tf.float64)

# Printing the input tensor
print('a: ', a)

# Calculating the result
res = tf.math.l2_normalize(x = a, axis = 1)

# Printing the result
print('Result: ', res)

Output:

a:  tf.Tensor(
[[ 7.  8.]
 [13. 11.]], shape=(2, 2), dtype=float64)
Result:  tf.Tensor(
[[0.65850461 0.75257669]
 [0.76338629 0.64594224]], shape=(2, 2), dtype=float64)