Source code for smac.acquisition.maximizer.local_and_random_search

from __future__ import annotations

from typing import Any

from ConfigSpace import Configuration, ConfigurationSpace

from smac.acquisition.function import AbstractAcquisitionFunction
from smac.acquisition.maximizer.abstract_acqusition_maximizer import (
    AbstractAcquisitionMaximizer,
)
from smac.acquisition.maximizer.local_search import LocalSearch
from smac.acquisition.maximizer.random_search import RandomSearch
from smac.utils.logging import get_logger

__copyright__ = "Copyright 2022, automl.org"
__license__ = "3-clause BSD"

logger = get_logger(__name__)


[docs]class LocalAndSortedRandomSearch(AbstractAcquisitionMaximizer): """Implement SMAC's default acquisition function optimization. This optimizer performs local search from the previous best points according, to the acquisition function, uses the acquisition function to sort randomly sampled configurations. Random configurations are interleaved by the main SMAC code. The Random configurations are interleaved to circumvent issues from a constant prediction from the Random Forest model at the beginning of the optimization process. Parameters ---------- configspace : ConfigurationSpace acquisition_function : AbstractAcquisitionFunction | None, defaults to None challengers : int, defaults to 5000 Number of challengers. max_steps: int | None, defaults to None [LocalSearch] Maximum number of steps that the local search will perform. n_steps_plateau_walk: int, defaults to 10 [LocalSearch] number of steps during a plateau walk before local search terminates. local_search_iterations: int, defauts to 10 [Local Search] number of local search iterations. seed : int, defaults to 0 """ def __init__( self, configspace: ConfigurationSpace, acquisition_function: AbstractAcquisitionFunction | None = None, challengers: int = 5000, max_steps: int | None = None, n_steps_plateau_walk: int = 10, local_search_iterations: int = 10, seed: int = 0, ) -> None: super().__init__( configspace, acquisition_function=acquisition_function, challengers=challengers, seed=seed, ) self._random_search = RandomSearch( configspace=configspace, acquisition_function=acquisition_function, seed=seed, ) self._local_search = LocalSearch( configspace=configspace, acquisition_function=acquisition_function, max_steps=max_steps, n_steps_plateau_walk=n_steps_plateau_walk, seed=seed, ) self._local_search_iterations = local_search_iterations @property def acquisition_function(self) -> AbstractAcquisitionFunction | None: # noqa: D102 """Returns the used acquisition function.""" return self._acquisition_function @acquisition_function.setter def acquisition_function(self, acquisition_function: AbstractAcquisitionFunction) -> None: self._acquisition_function = acquisition_function self._random_search._acquisition_function = acquisition_function self._local_search._acquisition_function = acquisition_function @property def meta(self) -> dict[str, Any]: # noqa: D102 meta = super().meta meta.update( { "random_search": self._random_search.meta, "local_search": self._local_search.meta, } ) return meta def _maximize( self, previous_configs: list[Configuration], n_points: int, ) -> list[tuple[float, Configuration]]: # Get configurations sorted by EI next_configs_by_random_search_sorted = self._random_search._maximize( previous_configs=previous_configs, n_points=n_points, _sorted=True, ) next_configs_by_local_search = self._local_search._maximize( previous_configs=previous_configs, n_points=self._local_search_iterations, additional_start_points=next_configs_by_random_search_sorted, ) # Having the configurations from random search, sorted by their # acquisition function value is important for the first few iterations # of SMAC. As long as the random forest predicts constant value, we # want to use only random configurations. Having them at the begging of # the list ensures this (even after adding the configurations by local # search, and then sorting them) next_configs_by_acq_value = next_configs_by_random_search_sorted + next_configs_by_local_search next_configs_by_acq_value.sort(reverse=True, key=lambda x: x[0]) first_five = [f"{_[0]} ({_[1].origin})" for _ in next_configs_by_acq_value[:5]] logger.debug(f"First 5 acquisition function values of selected configurations:\n{', '.join(first_five)}") return next_configs_by_acq_value
[docs]class LocalAndSortedPriorRandomSearch(AbstractAcquisitionMaximizer): """Implements SMAC's default acquisition function optimization. This optimizer performs local search from the previous best points according to the acquisition function, uses the acquisition function to sort randomly sampled configurations. Random configurations are interleaved by the main SMAC code. The random configurations are retrieved from two different ConfigurationSpaces - one which uses priors (e.g. NormalFloatHP) and is defined by the user, and one that is a uniform version of the same space, i.e. with the priors removed. Parameters ---------- configspace : ConfigurationSpace The original ConfigurationSpace specified by the user. uniform_configspace : ConfigurationSpace A version of the user-defined ConfigurationSpace where all parameters are uniform (or have their weights removed in the case of a categorical hyperparameter). acquisition_function : AbstractAcquisitionFunction | None, defaults to None challengers : int, defaults to 5000 Number of challengers. max_steps: int, defaults to None [LocalSearch] Maximum number of steps that the local search will perform. n_steps_plateau_walk: int, defaults to 10 [LocalSearch] number of steps during a plateau walk before local search terminates. local_search_iterations: int, defaults to 10 [Local Search] number of local search iterations. prior_sampling_fraction: float, defaults to 0.5 The ratio of random samples that are taken from the user-defined ConfigurationSpace, as opposed to the uniform version. seed : int, defaults to 0 """ def __init__( self, configspace: ConfigurationSpace, uniform_configspace: ConfigurationSpace, acquisition_function: AbstractAcquisitionFunction | None = None, challengers: int = 5000, max_steps: int | None = None, n_steps_plateau_walk: int = 10, local_search_iterations: int = 10, prior_sampling_fraction: float = 0.5, seed: int = 0, ) -> None: super().__init__( acquisition_function, configspace, challengers=challengers, seed=seed, ) self._prior_random_search = RandomSearch( acquisition_function=acquisition_function, configspace=configspace, seed=seed, ) self._uniform_random_search = RandomSearch( acquisition_function=acquisition_function, configspace=uniform_configspace, seed=seed, ) self._local_search = LocalSearch( acquisition_function=acquisition_function, configspace=configspace, max_steps=max_steps, n_steps_plateau_walk=n_steps_plateau_walk, seed=seed, ) self._local_search_iterations = local_search_iterations self._prior_sampling_fraction = prior_sampling_fraction def _maximize( self, previous_configs: list[Configuration], n_points: int, ) -> list[tuple[float, Configuration]]: # Get configurations sorted by EI next_configs_by_prior_random_search_sorted = self._prior_random_search._maximize( previous_configs, round(n_points * self._prior_sampling_fraction), _sorted=True, ) # Get configurations sorted by EI next_configs_by_uniform_random_search_sorted = self._uniform_random_search._maximize( previous_configs, round(n_points * (1 - self._prior_sampling_fraction)), _sorted=True, ) next_configs_by_random_search_sorted = [] next_configs_by_random_search_sorted.extend(next_configs_by_prior_random_search_sorted) next_configs_by_random_search_sorted.extend(next_configs_by_uniform_random_search_sorted) next_configs_by_local_search = self._local_search._maximize( previous_configs, self._local_search_iterations, additional_start_points=next_configs_by_random_search_sorted, ) # Having the configurations from random search, sorted by their # acquisition function value is important for the first few iterations # of SMAC. As long as the random forest predicts constant value, we # want to use only random configurations. Having them at the begging of # the list ensures this (even after adding the configurations by local # search, and then sorting them) next_configs_by_acq_value = next_configs_by_random_search_sorted + next_configs_by_local_search next_configs_by_acq_value.sort(reverse=True, key=lambda x: x[0]) logger.debug( "First 5 acq func (origin) values of selected configurations: %s", str([[_[0], _[1].origin] for _ in next_configs_by_acq_value[:5]]), ) return next_configs_by_acq_value