class DensityFilter[source]

DensityFilter(filter_size:int, dtype:dtype=torch.float32) :: Module

A parent class that inherits several different filters for smoothing.

Type Default Details
filter_size int The size of the filter.
dtype dtype torch.float32 The datatype of the filter.

DensityFilter.__call__[source]

DensityFilter.__call__(θ:Tensor)

Apply the filtering to the input. Returns a torch.Tensor.

Type Default Details
θ Tensor The input of the filter.

class MaxPoolDensityFilter[source]

MaxPoolDensityFilter(filter_size:int, dtype:dtype=torch.float32) :: DensityFilter

A filter that applies max pooling.

Type Default Details
filter_size int The size of the filter.
dtype dtype torch.float32 The datatype of the filter.

class ConvolutionDensityFilter[source]

ConvolutionDensityFilter(filter_size:int, dtype:dtype=torch.float32) :: DensityFilter

A parent class that inherits convolutional filters.

Type Default Details
filter_size int The size of the filter.
dtype dtype torch.float32 The datatype of the filter.

class UniformDensityFilter[source]

UniformDensityFilter(filter_size:int, dtype:dtype=torch.float32) :: ConvolutionDensityFilter

A class that performs convolution with a uniform filter, which is also refered to as mean pooling.

Type Default Details
filter_size int The size of the filter.
dtype dtype torch.float32 The datatype of the filter.

class RadialDensityFilter[source]

RadialDensityFilter(filter_size:int, dtype:dtype=torch.float32) :: ConvolutionDensityFilter

A class that performs convolution with a radial filter. A radial filter is a filter that has its maximal value in the center and decays radially to the outside. All values of the filter sum up to one.

Type Default Details
filter_size int The size of the filter.
dtype dtype torch.float32 The datatype of the filter.