Model.GaussianProcess

Gaussian process inference.

Definitions

data GaussianProcess : (0 _ : Shape) -> Type

  A Gaussian process is a collection of random variables, any finite number of which have joint
  Gaussian distribution. It can be viewed as a function from a feature space to a joint Gaussian
  distribution over a target space.
  
  @features The shape of the feature domain.

Totality: total
Visibility: public export
Constructor: 

MkGP : MeanFunction features -> Kernel features -> GaussianProcess features

  Construct a `GaussianProcess` as a pair of mean function and kernel.

data ConjugateGPRegression : (0 _ : Shape) -> Type

  A trainable model implementing vanilla Gaussian process regression. That is, regression with a
  Gaussian process as conjugate prior for homoscedastic Gaussian likelihoods. See the following
  for details:
  
  Gaussian Processes for Machine Learning
  Carl Edward Rasmussen and Christopher K. I. Williams
  The MIT Press, 2006. ISBN 0-262-18253-X.
  
  or
  
  Pattern Recognition and Machine Learning, Christopher M. Bishop

Totality: total
Visibility: public export
Constructor: 

MkConjugateGPR : (Tensor [p] F64 -> Tag (GaussianProcess features)) -> Tensor [p] F64 -> Tensor [] F64 -> ConjugateGPRegression features

  @gpFromHyperparameters Constructs a Gaussian process from the hyperparameters (presented as
    a vector)
  @hyperparameters The hyperparameters (excluding noise) presented as a vector.
  @noise The likehood amplitude, or observation noise.

fit : (Tensor [n] F64 -> Optimizer (Tensor [n] F64)) -> Dataset features [1] -> ConjugateGPRegression features -> Tag (ConjugateGPRegression features)

  Fit the Gaussian process and noise to the specified data.

Totality: total
Visibility: export