selector.generators.surrogates.smac_surrogate

This module contains functions for the SMAC surrogate.

Classes

SmacSurr(scenario[, seed, pca_dim])

Surrogate from SMAC.

class selector.generators.surrogates.smac_surrogate.SmacSurr(scenario, seed=False, pca_dim=8)

Bases: object

Surrogate from SMAC.

Note

Implementation is using source code of the package smac.

Parameters:

• scenario (selector.scenario.Scenario) – AC scenario.

• seed (int) – Random seed.

• pca_dim (int) – PCA dimension for SMAC surrogates instance feature PCA.

expected_improvement(suggestions, _)

Compute expected improvement via SMAC model.

Parameters:

• suggestions (list of selector.pool.Configuration) – Suggested configurations.

• _ (list of str) – List of next instances to be run.

Returns:

ei: Expected improvements.

Return type:

ndarray

get_suggestions(scenario, n_samples, *args)

Suggest configurations to run next based on the next instance set to run on.

Parameters:

• scenario (selector.scenario.Scenario) – AC scenario.

• n_samples (int) – Number of configurations to return.

• *args (Any) – Catches unneeded arguments due to the implementation of other surrogates.

Returns:

Suggested configurations.

Return type:

list of selector.pool.Configuration

pca_inst_feat_file(train_insts, feature_file, insts, pca_feats)

Generate instance feature file with PCA features.

Parameters:

• train_insts (str) – Name of instance file.

• feature_file (str) – path to file with problem instance features.

• insts (list of str) – Complete training instance set.

• pca_feats (ndarray) – Problem instance features.

Returns:

• ndarray, Instance features after PCA.

• list, Instance names.

Return type:

tuple

predict(confs, inst)

Predict performance/quality of configurations with GGA++ EPM.

Parameters:

• confs (list of selector.pool.Configuration) – Suggested configurations.

• inst (list) – List of next instances to run the tournament on.

Returns:

• ndarray, Mean of predicted performance/quality.

• ndarray, Variance of predicted performance/quality.

Return type:

tuple

probability_improvement(suggestions, results, i)

Compute probability of improvement.

Parameters:

• suggestions (list of selector.pool.Configuration) – Suggested configurations.

• results (dict) – Performances of the configuration on the instance set of the tournament.

• i (list) – List of next instances to run the tournament on.

Returns:

probimp: Probabilities of improvement.

Return type:

ndarray

transfom_selector_scenario_for_smac(scenario)

Transform scenario to SMAC formulation.

Parameters:

scenario (selector.scenario.Scenario) – AC scenario.

Returns:

• s : smac.scenario, AC scenario in SMAC format.

• config_space : smac.configspace.ConfigurationSpace, Parameter space definition in SMAC format.

• types : list, Parameter types.

• bounds : list of float, Parameter bounds.

Return type:

tuple

transform_for_epm(confs, pred=False)

Transform configuration to suit SMAC epm.

Parameters:

• confs (list selector.pool.Configuration) – Configurations to be transformed.

• pred (bool) – True if prepared for prediction.

Returns:

Transformed configurations.

Return type:

ndarray

transform_values(conf, pred=False)

Transform configuration values in SMAC format.

Parameters:

• conf (selector.pool.Configuration) – Configuration to be transformed.

• pred (bool) – True if configuration is prepared for prediction.

Returns:

Transformed configuration values.

Return type:

dict

update(history, configs, results, terminations, ac_runtime=None)

Update SMAC epm.

Parameters:

• history (list of selector.pool.Tournament) – Tournament history.

• configs (list of selector.pool.Configuration) – Configurations that participated in the tournament.

• results (dict) – Results of the tournament.

• terminations (dict) – Information about terminations of runs that occurred.

• ac_runtime (int) – Total AC runtime in seconds so far.