Skip to contents

Abstract base class for distance-aware active-learning acquisition functions.

The acquisition constants include a detached copy of the wrapped ALDistance parameter set under the al_distance prefix. During update and evaluation, those values are copied into the distance before reference points are set or distances are computed.

$fit_pool() fits the distance on the candidate pool used for scaling and pool-derived acquisition state, or on the search space when xdt is NULL. $update() then sets the distance reference points to the current archive rows in archive order.

Creates a new distance-aware acquisition function.

Fits the wrapped distance and any subclass pool state on a finite candidate pool or on the search space.

Sets the distance reference points to the current surrogate archive.

Arguments

id

(character(1))
Acquisition function id.

constants

(paradox::ParamSet)
Changeable constants or parameters of the acquisition function.

surrogate

(NULL | mlr3mbo::Surrogate)
Surrogate whose predictions are used in the acquisition function.

al_distance

(NULL | ALDistance)
Distance object used by the acquisition function.

requires_predict_type_se

(logical(1))
Whether the surrogate must use predict type "se".

surrogate_class

(character(1))
Allowed class of the surrogate. Default "Surrogate".

direction

("same" | "minimize" | "maximize")
Optimization direction of the acquisition function.

packages

(character())
Required packages.

label

(character(1))
Label for the object.

man

(character(1))
String in the format [pkg]::[topic] pointing to a manual page.

xdt

(NULL | data.table)
Candidate configurations in search-space coordinates. NULL fits to the search space where supported.

search_space

(paradox::ParamSet)
Search space describing the candidate-pool columns.

Value

self.

Fields

al_distance

(NULL | ALDistance) Distance object used by the acquisition function.