API Reference

Bayes-skopt, or bask, builds on Scikit-Optimize and implements a fully Bayesian sequential optimization framework of very noise black-box functions.

bask: module

Base classes

BayesGPR([kernel, alpha, optimizer, …])

Gaussian process regressor of which the kernel hyperparameters are inferred in a fully Bayesian framework.

BayesSearchCV(estimator, search_spaces[, …])

Fully Bayesian optimization over hyper parameters.

Optimizer(dimensions[, n_points, …])

Execute a stepwise Bayesian optimization.

Functions

guess_priors(kernel)

Guess suitable priors for the hyperparameters of a given kernel.

r2_sequence(n, d[, seed])

Output n points of the infinite R2 quasi-random sequence.

bask.acquisition: Acquisition

User guide: See the acquisition section for further details.

acquisition.PVRS()

Implements the predictive variance reduction search algorithm.

acquisition.MaxValueSearch()

Select points based on their mutual information with the optimum value.

acquisition.ExpectedImprovement()

Select the point maximizing the expected improvement over the current optimum.

acquisition.TopTwoEI()

Select the point with the highest expected improvement over the point with the maximum expected improvement overall.

acquisition.LCB()

Select the point with the lowest lower confidence bound.

acquisition.Expectation()

Select the point with the lowest estimated mean.

acquisition.ThompsonSampling()

Sample a random function from the GP and select its optimum.

acquisition.VarianceReduction()

A criterion which tries to find the region where it can reduce the global variance the most.

acquisition.evaluate_acquisitions(X, gpr[, …])

Run a set of acquisitions functions on a given set of points.

bask.optimizer: Optimizer

User guide: See the optimizer section for further details.

optimizer.Optimizer(dimensions[, n_points, …])

Execute a stepwise Bayesian optimization.

bask.utils: Utils functions.

User guide: See the utils section for further details.

utils.geometric_median(X[, eps])

Compute the geometric median for the given array of points.

utils.r2_sequence(n, d[, seed])

Output n points of the infinite R2 quasi-random sequence.

utils.guess_priors(kernel)

Guess suitable priors for the hyperparameters of a given kernel.

utils.construct_default_kernel(dimensions)

Construct a Matern kernel as default kernel to be used in the optimizer.