In this post, we will be seeing how to initialize a vector in TensorFlow with all zeros or ones. The function you will be calling is tf.ones(). To initialize with zeros you could use tf.zeros() instead. These functions take in a shape and return an array full of zeros and ones accordingly.
Code:
import tensorflow as tf ones_matrix = tf.ones([2, 3]) sess = tf.Session() ones = sess.run(ones_matrix) sess.close() print(ones) |
Output:
[[1. 1. 1.] [1. 1. 1.]]
Using One Hot Encoding:
Many times in deep learning and general vector computations you will have a y vector with numbers ranging from 0 to C-1 and you want to do the following conversion. If C is for example 5, then you might have the following y vector which you will need to convert as follows:
One Hot Encoding Example
This can be done as follows:
Parameters passed to the function:
indices: A Tensor of indices.
depth: A scalar defining the depth of the one hot dimension.
on_value: A scalar defining the value to fill in output when indices[j] = i. (default : 1)
off_value: A scalar defining the value to fill in output when indices[j] != i. (default : 0)
axis: The axis to fill (default : -1, a new inner-most axis).
dtype: The data type of the output tensor.
name: A name for the operation (optional).
Code:
indices = [1, 4, 2, 0, 3] C = tf.constant(5, name = "C") one_hot_matrix = tf.one_hot( indices, C, on_value = 1.0, off_value = 0.0, axis =-1) sess = tf.Session() one_hot = sess.run(one_hot_matrix) sess.close() # output is of dimension 5 x 5 print(one_hot) |
Output:
[[0.0, 1.0, 0.0, 0.0, 0.0 ] [0.0, 0.0, 0.0, 0.0, 1.0] [0.0, 0.0, 1.0, 0.0, 0.0] [1.0, 0.0, 0.0, 0.0, 0.0] [0.0, 0.0, 0.0, 1.0, 0.0]]
Feel free to change values and see the result.
Code:
indices = [[0, 2], [1, -1]] C = tf.constant(5, name = "C") one_hot_matrix = tf.one_hot( indices, C, on_value = 1.0, off_value = 0.0, axis =-1) sess = tf.Session() one_hot = sess.run(one_hot_matrix) sess.close() # output is of dimension 2 x 2 x 3 print(one_hot) |
Output :
[[[1.0, 0.0, 0.0], [0.0, 0.0, 1.0]], [[0.0, 1.0, 0.0], [0.0, 0.0, 0.0]]]
