Tensorflow: list of index -> generate an array

Question

I have a TensorFlow question about building TF array given a list of indices of its elements.

Say I have an index list with tf array

list_index_false = tf.constant([5,3])

I would like to build size 7 tf.array with booleans where only indices 5 and 3 are False while others are True such as below:
[True,True,True,False,True,False,True]

I tried following:

list_boolean=tf.fill([7],True))

Then tried to assign list_boolean[3]=False, list_boolean[5]=False, but tensorflow doesn't let me to assign. :(

How can I do that? Any other way?

Thank you

Max · Accepted Answer · 2017-12-24 07:22:55Z

1

tf.scatter_update is specifically designed for usecases like this. Here is the code example:

import tensorflow as tf

list_index_false = tf.constant([5, 3])
list_boolean = tf.Variable(tf.fill([7], True))
list_boolean = tf.scatter_update(list_boolean, list_index_false, tf.fill(tf.shape(list_index_false), False))

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    print(sess.run(list_boolean))

Hope it helps!

answered Dec 24, 2017 at 7:22

Max

1,0396 silver badges7 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

JimSD Over a year ago

How can i possibly thank you! I have been in trouble for a week startin to hate Tensorflow.. You saved me.!

Max Over a year ago

Glad to help :)

Collectives™ on Stack Overflow

Tensorflow: list of index -> generate an array

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related