torchadf.nn.functional.conv2d

torchadf.nn.functional.conv2d(in_mean, in_var, weight, bias=None, stride=1, padding=0, dilation=1, groups=1, mode='diag')

Applies a convolution function for 2D inputs.

Assumed Density Filtering (ADF) version of torch.nn.functional.conv2d.

Parameters:

in_meantorch.Tensor

Input mean tensor.

in_vartorch.Tensor

Input (co-)variance tensor.

weighttorch.Tensor or torch.nn.parameter.Parameter

The convolution filter weights.

biastorch.Tensor or torch.nn.parameter.Parameter, optional

The additive convolution bias (Default None).

strideint or tuple of int, optional

Stride of the convolution kernel (Default 1).

paddingstr or int or tuple of int, optional

Implicit padding on both sides of input (Default 0).

dilationint or tuple of int, optional

Spacing between kernel elements (Default 1).

groupsint, optional

Split convolution into a number independent groups (Default 1). Number of input channels has to be divisible by the number of groups.

mode{“diag”, “diagonal”, “lowrank”, “half”, “full”}, optional

Covariance propagation mode (Default “diag”).

Returns:

out_meantorch.Tensor

The transformed mean tensor.

out_vartorch.Tensor

The transformed (co-)variance tensor.