Crate egobox_gp

This library implements Gaussian Process regression also known as Kriging models, it is a port of SMT Kriging and KPLS surrogate models.

It also implements Sparse Gaussian Processes methods (SGPs) which address limitations of Gaussian Processes (GPs) when the number of training points is large. Indeed the complexity of GPs algorithm is in O(N^3) in processing time and O(N^2) in memory where N is the number of training points. The complexity is then respectively reduced to O(N.M^2) and O(NM) where M is the number of so-called inducing points with M < N.

GP methods are implemented by GaussianProcess parameterized by GpParams.

SGP methods are implemented by SparseGaussianProcess parameterized by SgpParams.

Modules

• correlation_models
  A module for correlation models with PLS weighting to model the error term of the GP model.
• mean_models
  A module for regression models to model the mean term of the GP model. In practice small degree (<= 2) polynomial regression models are used, as the gaussian process is then fitted using the correlated error term.

Structs

• GaussianProcess
  A GP regression is an interpolation method where the interpolated values are modeled by a Gaussian process with a mean and governed by a prior covariance kernel, which depends on some parameters to be determined.
• GpParams
  The set of hyperparameters that can be specified for the execution of the GP algorithm.
• GpValidParams
  A set of validated GP parameters.
• ParamTuning
  A structure to represent a n-dim parameter estimation
• SgpParams
  The set of hyperparameters that can be specified for the execution of the SGP algorithm.
• SgpValidParams
  A set of validated SGP parameters.
• SparseGaussianProcess
  Sparse gaussian process considers a set of M inducing points either to approximate the posterior Gaussian distribution with a low-rank representation (FITC - Fully Independent Training Conditional method), or to approximate the posterior distribution directly (VFE - Variational Free Energy method).
• ThetaTuning
  As structure for theta hyperparameters guess

Enums

• GpError
  An error when using GaussianProcess or a SparseGaussianProcess algorithm
• Inducings
  SGP inducing points specification
• ParamEstimation
  Variance estimation method
• SparseMethod
  SGP algorithm method specification

Type Aliases

• Kriging
  Kriging as GP special case when using constant mean and squared exponential correlation
• Result
  A result type for GP regression algorithm
• SparseKriging
  Kriging as sparse GP special case when using squared exponential correlation