tf.contrib.framework.convolutional_orthogonal_3d

Class convolutional_orthogonal_3d

Defined in tensorflow/python/ops/init_ops.py.

Initializer that generates a 3D orthogonal kernel for ConvNets.

The shape of the tensor must have length 5. The number of input filters must not exceed the number of output filters. The orthogonality(==isometry) is exact when the inputs are circular padded. There are finite-width effects with non-circular padding (e.g. zero padding). See algorithm 1 [Xiao et al., 2018] in: https://arxiv.org/abs/1806.05393

Args:

  • gain: Multiplicative factor to apply to the orthogonal matrix. Default is 1. The 2-norm of an input is multiplied by a factor of 'sqrt(gain)' after applying this convolution.
  • seed: A Python integer. Used to create random seeds. See tf.set_random_seed for behavior.
  • dtype: The data type.

Methods

__init__

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

__call__

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

from_config

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

get_config()