LaplaceConvNd

class torch_uncertainty.layers.distributions.LaplaceConvNd(base_layer, event_dim, min_scale=1e-06, **layer_args)

Laplace Distribution Convolutional Density Layer.

Parameters:

• in_channels (int) – The number of input channels.

• out_channels (int) – The number of event channels.

• kernel_size (int | tuple[int]) – The size of the convolutional kernel.

• stride (int | tuple[int]) – The stride of the convolution.

• padding (int | tuple[int]) – The padding of the convolution.

• dilation (int | tuple[int]) – The dilation of the convolution.

• groups (int) – The number of groups in the convolution.

• min_scale (float) – The minimal value of the scale parameter.

• device (torch.device) – The device where the layer is stored.

• dtype (torch.dtype) – The datatype of the layer.

Shape:

• Input: \((N, C_{in}, \ast)\) where \(\ast\) means any number of dimensions and \(C_{in} = \text{in_channels}\) and \(N\) is the batch size.

• Output: A dict with the following keys

  • "loc": The mean of the Laplace distribution of shape \((N, C_{out}, \ast)\) where \(C_{out} = \text{out_channels}\).

  • "scale": The standard deviation of the Laplace distribution of shape \((\ast, C_{out}, \ast)\).