tensorflow boolean_ Mask function

The function prototype

tf.boolean_mask(
    tensor, mask, axis=None, name='boolean_mask'
)

Function description

For preserving tensors tenso Part of , It depends on the tensor mask The subscript .

Parameters tensor Represents the tensor to be manipulated , Parameters mask Is a Boolean matrix , decision tensor Those places should be preserved , Parameters axis Indicates which axis to start from .

Function USES

1、 One dimensional tensor

>>> tensor = tf.constant([1, 2, 3, 4])
>>> mask = tf.constant([True, True, False, False])
>>> tf.boolean_mask(tensor, mask)
<tf.Tensor: shape=(2,), dtype=int32, numpy=array([1, 2])>

2、 Two dimensional tensor ,mask If it is a one-dimensional tensor , You need to be with axis The length of the corresponding axis is equal , If it is a two-dimensional tensor , You need to be with tensor Of shape identical .

>>> tensor = tf.constant([[1, 2], [3, 4]])
>>> mask = tf.constant([True, False])
#  The default from the axis=0 Start 
>>> tf.boolean_mask(tensor, mask)
<tf.Tensor: shape=(1, 2), dtype=int32, numpy=array([[1, 2]])>

>>> tensor = tf.constant([[1, 2], [3, 4]])
>>> mask = tf.constant([True, False])
>>> tf.boolean_mask(tensor, mask, axis=1)
<tf.Tensor: shape=(2, 1), dtype=int32, numpy=
array([[1],
       [3]])>

>>> tensor = tf.constant([[1, 2], [3, 4]])
>>> mask = tf.constant([[True, True], [True, False]])
>>> tf.boolean_mask(tensor, mask)
<tf.Tensor: shape=(3,), dtype=int32, numpy=array([1, 2, 3])>