C++ API Reference for Intel® Data Analytics Acceleration Library 2020 Update 1

Namespaces | Classes | Enumerations
daal::algorithms::neural_networks::layers::transposed_conv2d Namespace Reference

Contains classes for neural network 2D transposed convolution layer.

Namespaces

 backward
 Contains classes for the backward 2D transposed convolution layer.
 
 forward
 Contains classes for the forward 2D transposed convolution layer.
 
 interface1
 Contains version 1.0 of Intel(R) Data Analytics Acceleration Library (Intel(R) DAAL) interface.
 

Classes

struct  KernelSizes
 Data structure representing the size of the two-dimensional kernel subtensor. More...
 
struct  Strides
 Data structure representing the intervals on which the subtensors for 2D transposed convolution are selected. More...
 
struct  Paddings
 Data structure representing the number of data elements to implicitly add to each size of the two-dimensional subtensor on which 2D transposed convolution is performed. More...
 
struct  Indices
 Data structure representing the indices of the two dimensions on which 2D transposed convolution is performed. More...
 
struct  ValueSizes
 Data structure representing the value sizes of the two dimensions on which 2D transposed convolution is performed. More...
 

Enumerations

enum  Method { defaultDense = 0 }
 
enum  LayerDataId { auxData, auxWeights }
 

Enumeration Type Documentation

Available identifiers of results of the forward 2D transposed convolution layer and input objects for the backward 2D transposed convolution layer

Enumerator
auxData 

Data processed at the forward stage of the layer

auxWeights 

Input weights for forward stage of the layer

enum Method

Available methods to compute forward and backward 2D transposed convolution layer

Enumerator
defaultDense 

Default: performance-oriented method.

For more complete information about compiler optimizations, see our Optimization Notice.