tf.nn.dropout

Computes dropout: randomly sets elements to zero to prevent overfitting.

tf.nn.dropout(
    x, rate, noise_shape=None, seed=None, name=None
)

Used in the notebooks

Used in the guide

See also: tf.keras.layers.Dropout for a dropout layer.

Dropout is useful for regularizing DNN models. Inputs elements are randomly set to zero (and the other elements are rescaled). This encourages each node to be independently useful, as it cannot rely on the output of other nodes.

More precisely: With probability rate elements of x are set to 0. The remaining elements are scaled up by 1.0 / (1 - rate), so that the expected value is preserved.

tf.random.set_seed(0)
x = tf.ones([3,5])
tf.nn.dropout(x, rate = 0.5, seed = 1).numpy()
array([[2., 0., 0., 2., 2.],
     [2., 2., 2., 2., 2.],
     [2., 0., 2., 0., 2.]], dtype=float32)
 
tf.random.set_seed(0)
x = tf.ones([3,5])
tf.nn.dropout(x, rate = 0.8, seed = 1).numpy()
array([[0., 0., 0., 5., 5.],
     [0., 5., 0., 5., 0.],
     [5., 0., 5., 0., 5.]], dtype=float32)
 
tf.nn.dropout(x, rate = 0.0) == x
<tf.Tensor: shape=(3, 5), dtype=bool, numpy=
  array([[ True,  True,  True,  True,  True],
         [ True,  True,  True,  True,  True],
         [ True,  True,  True,  True,  True]])>

By default, each element is kept or dropped independently. If noise_shape is specified, it must be broadcastable to the shape of x, and only dimensions with noise_shape[i] == shape(x)[i] will make independent decisions. This is useful for dropping whole channels from an image or sequence. For example:

tf.random.set_seed(0)
x = tf.ones([3,10])
tf.nn.dropout(x, rate = 2/3, noise_shape=[1,10], seed=1).numpy()
array([[0., 0., 0., 3., 3., 0., 3., 3., 3., 0.],
     [0., 0., 0., 3., 3., 0., 3., 3., 3., 0.],
     [0., 0., 0., 3., 3., 0., 3., 3., 3., 0.]], dtype=float32)

x A floating point tensor.
rate A scalar Tensor with the same type as x. The probability that each element is dropped. For example, setting rate=0.1 would drop 10% of input elements.
noise_shape A 1-D integer Tensor, representing the shape for randomly generated keep/drop flags.
seed A Python integer. Used to create random seeds. See tf.random.set_seed for behavior.
name A name for this operation (optional).

A Tensor of the same shape of x.

ValueError If rate is not in [0, 1) or if x is not a floating point tensor. rate=1 is disallowed, because the output would be all zeros, which is likely not what was intended.