The transpose of
tf.nn.conv3d_transpose( value, filter=None, output_shape=None, strides=None, padding='SAME', data_format='NDHWC', name=None, input=None, filters=None, dilations=None )
This operation is sometimes called "deconvolution" after Deconvolutional
but is really the transpose (gradient) of
conv3d rather than an actual
value: A 5-D
[batch, depth, height, width, in_channels].
filter: A 5-D
Tensorwith the same type as
[depth, height, width, output_channels, in_channels].
in_channelsdimension must match that of
output_shape: A 1-D
Tensorrepresenting the output shape of the deconvolution op.
strides: A list of ints. The stride of the sliding window for each dimension of the input tensor.
padding: A string, either
'SAME'. The padding algorithm. See the "returns" section of
data_format: A string, either
'NCDHW' specifying the layout of the input and output tensors. Defaults to
name: Optional name for the returned tensor.
input: Alias of value.
filters: Alias of filter.
dilations: An int or list of
intsthat has length
5, defaults to 1. The dilation factor for each dimension of
input. If a single value is given it is replicated in the
Wdimension. By default the
Cdimensions are set to 1. If set to k > 1, there will be k-1 skipped cells between each filter element on that dimension. The dimension order is determined by the value of
data_format, see above for details. Dilations in the batch and depth dimensions if a 5-d tensor must be 1.
Tensor with the same type as
ValueError: If input/output depth does not match
filter's shape, or if padding is other than