Missed TensorFlow World? Check out the recap. Learn more

tf.initializers.uniform_unit_scaling

View source on GitHub

Class uniform_unit_scaling

Initializer that generates tensors without scaling variance.

Inherits From: Initializer

Aliases:

  • Class tf.compat.v1.initializers.uniform_unit_scaling
  • Class tf.compat.v1.uniform_unit_scaling_initializer
  • Class tf.uniform_unit_scaling_initializer

When initializing a deep network, it is in principle advantageous to keep the scale of the input variance constant, so it does not explode or diminish by reaching the final layer. If the input is x and the operation x * W, and we want to initialize W uniformly at random, we need to pick W from

[-sqrt(3) / sqrt(dim), sqrt(3) / sqrt(dim)]

to keep the scale intact, where dim = W.shape[0] (the size of the input). A similar calculation for convolutional networks gives an analogous result with dim equal to the product of the first 3 dimensions. When nonlinearities are present, we need to multiply this by a constant factor. See (Sussillo et al., 2014) for deeper motivation, experiments and the calculation of constants. In section 2.3 there, the constants were numerically computed: for a linear layer it's 1.0, relu: ~1.43, tanh: ~1.15.

Args:

  • factor: Float. A multiplicative factor by which the values will be scaled.
  • seed: A Python integer. Used to create random seeds. See tf.compat.v1.set_random_seed for behavior.
  • dtype: Default data type, used if no dtype argument is provided when calling the initializer. Only floating point types are supported.

References:

Sussillo et al., 2014 (pdf)

__init__

View source

__init__(
    factor=1.0,
    seed=None,
    dtype=tf.dtypes.float32
)

DEPRECATED FUNCTION (deprecated arguments)

Methods

__call__

View source

__call__(
    shape,
    dtype=None,
    partition_info=None
)

from_config

View source

from_config(
    cls,
    config
)

Instantiates an initializer from a configuration dictionary.

Example:

initializer = RandomUniform(-1, 1)
config = initializer.get_config()
initializer = RandomUniform.from_config(config)

Args:

  • config: A Python dictionary. It will typically be the output of get_config.

Returns:

An Initializer instance.

get_config

View source

get_config()