Eigen Namespace Reference

Namespaces
namespace	internal

Typedefs
typedef IndexList< type2index< 0 >, type2index< 0 >, type2index< 1 >, type2index< 1 > >	ReverseColMajor
	Computes the backprop for the input of a 2D convolution.
typedef IndexList< type2index< 1 >, type2index< 1 >, type2index< 0 >, type2index< 0 > >	ReverseRowMajor

Typedef Documentation

ReverseColMajor

typedef IndexList<type2index<0>, type2index<0>, type2index<1>, type2index<1> > Eigen::ReverseColMajor

Computes the backprop for the input of a 2D convolution.

SpatialConvolutionBackwardInput

The output_backward parameter is expected to be a tensor with a rank of 3 or more (channels, height, width, and optionally others) The kernel parameter is expected to be a 4D tensor (filters, channels, kernel_height, kernel_width) The output_backward and the kernel must both be in col-major layout. The result will also be in col-major layout.

If row_in_stride, col_in_stride > 1, then applies convolution with holes (aka atrous convolution), sampling every row_in_stride, col_in_stride input pixels.

The result can be assigned to a tensor of rank equal to the rank of the output_backward. The dimensions of the result will be filters, height, width (and others if applicable).

It is possible to swap the order of the width and height dimensions provided that the same order is used in the input, the kernel, and the output.

Definition at line 51 of file eigen_backward_spatial_convolutions.h.

ReverseRowMajor

typedef IndexList<type2index<1>, type2index<1>, type2index<0>, type2index<0> > Eigen::ReverseRowMajor

Definition at line 52 of file eigen_backward_spatial_convolutions.h.