RandomEMAssignment#

class sdm.prior.RandomEMAssignment(p: HP[float] | float, q: HP[float] | float, bandwidth: HP[float] | float | None = None, m: HP[int] | int | None = None, linear: RandomLinear | None = None)#

Bases: RandomFunction

Compute random soft cluster assignments and project them.

Parameters:
  • p (HP[float] | float) – Minkowski distance exponent.

  • q (HP[float] | float) – Exponent applied to the normalized distance.

  • bandwidth (HP[float] | float | None) – Log-scale spread of cluster widths; defaults to a correlated uniform draw between zero and one.

  • m (HP[int] | int | None) – Number of cluster centers. Defaults to LogUniformInt(2, max(16, 2 * out_channels)).

  • linear (RandomLinear | None) – Random linear function applied after softmax. Defaults to RandomLinear.

forward(x: Tensor, out_channels: int, generator: Generator | None = None) → Tensor#

Apply a sampled function to every row of x.

Parameters:
  • x (Tensor) – Input values with shape [R, H] for rows and channels.

  • out_channels (int) – Number of output channels.

  • generator (Generator | None) – Pseudorandom generator.

Returns:

Node values with shape [R, out_channels].

Return type:

Tensor