Developer Guide for Intel® Data Analytics Acceleration Library 2018 Update 3
The forward two-dimensional transposed convolution layer accepts the input described below. Pass the Input ID as a parameter to the methods that provide input for your algorithm. For more details, see Algorithms.
Input ID |
Input |
|
---|---|---|
data |
Pointer to tensor X of size n1 × l2 × l3 × l4 that stores the input data of the forward two-dimensional transposed convolution layer. This input can be an object of any class derived from Tensor. |
|
weighs |
Pointer to the tensor K of size l2 × nKernels/nGroups × m3 × m4 that stores a set of kernel weights. This input can be an object of any class derived from Tensor. |
|
biases |
Pointer to the tensor B of size nKernels that stores a set of biases. This input can be an object of any class derived from Tensor. If you pass a null pointer, the library does not apply a bias-related transformation. |
For common parameters of neural network layers, see Common Parameters.
In addition to the common parameters, the forward two-dimensional transposed convolution layer has the following parameters:
Parameter |
Default Value |
Description |
|
---|---|---|---|
algorithmFPType |
float |
The floating-point type that the algorithm uses for intermediate computations. Can be float or double. |
|
method |
defaultDense |
Performance-oriented computation method, the only method supported by the layer. |
|
kernelSizes |
KernelSizes(2, 2) |
Data structure representing the sizes mi, i ∈ {3, 4}, of the two-dimensional kernel subtensor. |
|
indices |
Indices(2,3) |
Data structure representing the dimensions for applying transposed convolution kernels. |
|
strides |
Strides(2, 2) |
Data structure representing the intervals si, i ∈ {3, 4}, on which the kernel should be applied to the input. |
|
paddings |
paddings(0, 0) |
Data structure representing the number of data elements pi, i ∈ {3, 4}, to implicitly add to each side of the two-dimensional subtensor along which forward two-dimensional transposed convolution is performed. |
|
valueSizes |
ValueSizes(0, 0) |
Data structure representing the dimension sizes ni, i ∈ {3, 4}, of the value tensor Y. If this value contains (0, 0), the size of value dimensions are computed automatically. |
|
nKernels |
n/a |
Number of kernels applied to the layer input data. |
|
groupDimension |
1 |
Dimension to which grouping is applied. |
|
nGroups |
1 |
Number of groups into which the input data is split in dimension groupDimension. |
The forward two-dimensional transposed convolution layer calculates the result described below. Pass the Result ID as a parameter to the methods that access the results of your algorithm. For more details, see Algorithms.
Result ID |
Result |
|
---|---|---|
value |
Pointer to tensor Y of size n1 x nKernels x n3 x n4 that stores the result of the forward two-dimensional transposed convolution layer. This result can be an object of any class derived from Tensor. |
|
resutForBackward |
Collection of data obtained on the forward two-dimensional transposed convolution layer. This collection can contain objects of any class derived from Tensor. | |
Element ID |
Element |
|
auxData |
Pointer to tensor X of size n1 × l2 × l3 × l4 that stores the input data for the forward two-dimensional transposed convolution layer. This result can be an object of any class derived from Tensor. |
|
auxWeights |
Pointer to tensor K of size l2 × nKernels/nGroups × m3 × m4 that stores a set of kernel weights. This result can be an object of any class derived from Tensor. |