Spidr.Model

Functionality for probabilistic models.

Definitions

interface ProbabilisticModel : Shape -> Shape -> ((0 _ : Shape) -> (0 _ : Nat) -> Type) -> Type -> Type

  A `ProbabilisticModel` is a mapping from a feature domain to a probability distribution over
  a target domain.
  
  @features The shape of the feature domain.
  @targets The shape of the target domain.
  @marginal The type of mulitvariate marginal distribution over the target domain.

Parameters: features, targets, marginal, model
Constraints: Distribution marginal
Methods:

marginalise : model -> Tensor (S n :: features) F64 -> Tag (marginal targets (S n))

  Return the marginal distribution over the target domain at the specified feature values.

marginalise : ProbabilisticModel features targets marginal model => model -> Tensor (S n :: features) F64 -> Tag (marginal targets (S n))

  Return the marginal distribution over the target domain at the specified feature values.

Totality: total
Visibility: public export