tf.sparse.segment_sum  |  TensorFlow v2.16.1 (original) (raw)

tf.sparse.segment_sum

Stay organized with collections Save and categorize content based on your preferences.

Computes the sum along sparse segments of a tensor.

tf.sparse.segment_sum(
    data,
    indices,
    segment_ids,
    num_segments=None,
    name=None,
    sparse_gradient=False
)

Read the section on segmentationfor an explanation of segments.

Like tf.math.segment_sum, but segment_ids can have rank less than data's first dimension, selecting a subset of dimension 0, specified by indices.segment_ids is allowed to have missing ids, in which case the output will be zeros at those indices. In those cases num_segments is used to determine the size of the output.

For example:

c = tf.constant([[1,2,3,4], [-1,-2,-3,-4], [5,6,7,8]])

# Select two rows, one segment.
tf.sparse.segment_sum(c, tf.constant([0, 1]), tf.constant([0, 0]))
# => [[0 0 0 0]]

# Select two rows, two segment.
tf.sparse.segment_sum(c, tf.constant([0, 1]), tf.constant([0, 1]))
# => [[ 1  2  3  4]
#     [-1 -2 -3 -4]]

# With missing segment ids.
tf.sparse.segment_sum(c, tf.constant([0, 1]), tf.constant([0, 2]),
                      num_segments=4)
# => [[ 1  2  3  4]
#     [ 0  0  0  0]
#     [-1 -2 -3 -4]
#     [ 0  0  0  0]]

# Select all rows, two segments.
tf.sparse.segment_sum(c, tf.constant([0, 1, 2]), tf.constant([0, 0, 1]))
# => [[0 0 0 0]
#     [5 6 7 8]]

# Which is equivalent to:
tf.math.segment_sum(c, tf.constant([0, 0, 1]))