bolero.behavior_search.ContextualBlackBoxSearch¶

class bolero.behavior_search.ContextualBlackBoxSearch(behavior, optimizer, metaparameter_keys=[], metaparameter_values=[])[source]¶

Combine a contextual black box optimizer with a black box behavior.

Black box in this context means that only a fixed number of parameters will optimized with respect to a scalar reward function.

Parameters:

behavior : dict or Behavior subclass A black-box behavior that is given directly or fully specified by a configuration dictionary. optimizer : dict or ContextualOptimizer subclass A black-box optimizer that is given directly or fully specified by a configuration dictionary. metaparameter_keys : list, optional (default: []) Names of metaparameters for the behavior that will be set during initialization. metaparameter_values : list, optional (default: []) Values of metaparameters for the behavior that will be set during initialization.

__init__(behavior, optimizer, metaparameter_keys=[], metaparameter_values=[])[source]¶

get_args()¶

Get parameters for this estimator.

Returns:	params : mapping of string to any Parameter names mapped to their values.

get_best_behavior_template()[source]¶

Return current best estimate of contextual policy.

get_desired_context()[source]¶

Chooses desired context for next evaluation.

Returns:	context : ndarray-like, default=None The context in which the next rollout shall be performed. If None, the environment may select the next context without any preferences.

set_context(context)[source]¶

Set context of next evaluation.

Note that the set context need not necessarily be the same that was requested by get_desired_context().

Parameters:	context : array-like, shape (n_context_dims,) The context in which the next rollout will be performed