def __init__(self, interface, **kwargs): super(NelderMeadController, self).__init__(interface, **kwargs) self.learner = mll.NelderMeadLearner( start_datetime=self.start_datetime, **self.remaining_kwargs) self._update_controller_with_learner_attributes() self.out_type.append('nelder_mead')
def __init__( self, interface, training_type='differential_evolution', machine_learner_type='machine_learner', num_training_runs=None, no_delay=True, num_params=None, min_boundary=None, max_boundary=None, trust_region=None, learner_archive_filename=mll.default_learner_archive_filename, learner_archive_file_type=mll.default_learner_archive_file_type, param_names=None, **kwargs): super(MachineLearnerController, self).__init__(interface, **kwargs) self.machine_learner_type = machine_learner_type self.last_training_cost = None self.last_training_bad = None self.last_training_run_flag = False if num_training_runs is None: if num_params is None: self.num_training_runs = 10 else: self.num_training_runs = max(10, 2 * int(num_params)) else: self.num_training_runs = int(num_training_runs) if self.num_training_runs <= 0: self.log.error( 'Number of training runs must be larger than zero:' + repr(self.num_training_runs)) raise ValueError self.no_delay = bool(no_delay) self.training_type = str(training_type) if self.training_type == 'random': self.learner = mll.RandomLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, trust_region=trust_region, learner_archive_filename=None, learner_archive_file_type=learner_archive_file_type, param_names=param_names, **self.remaining_kwargs) elif self.training_type == 'nelder_mead': self.learner = mll.NelderMeadLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, learner_archive_filename=None, learner_archive_file_type=learner_archive_file_type, param_names=param_names, **self.remaining_kwargs) elif self.training_type == 'differential_evolution': self.learner = mll.DifferentialEvolutionLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, trust_region=trust_region, evolution_strategy='rand2', learner_archive_filename=None, learner_archive_file_type=learner_archive_file_type, param_names=param_names, **self.remaining_kwargs) else: self.log.error( 'Unknown training type provided to machine learning controller:' + repr(training_type)) self.archive_dict.update({'training_type': self.training_type}) self._update_controller_with_learner_attributes()
def __init__( self, interface, training_type='random', num_training_runs=None, no_delay=True, num_params=None, min_boundary=None, max_boundary=None, trust_region=None, learner_archive_filename=mll.default_learner_archive_filename, learner_archive_file_type=mll.default_learner_archive_file_type, **kwargs): super(GaussianProcessController, self).__init__(interface, **kwargs) self.last_training_cost = None self.last_training_bad = None self.last_training_run_flag = False if num_training_runs is None: if num_params is None: self.num_training_runs = 10 else: self.num_training_runs = max(10, 2 * int(num_params)) else: self.num_training_runs = int(num_training_runs) if self.num_training_runs <= 0: self.log.error( 'Number of training runs must be larger than zero:' + repr(self.num_training_runs)) raise ValueError self.no_delay = bool(no_delay) self.training_type = str(training_type) if self.training_type == 'random': self.learner = mll.RandomLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, trust_region=trust_region, learner_archive_filename=None, learner_archive_file_type=learner_archive_file_type, **self.remaining_kwargs) elif self.training_type == 'nelder_mead': self.learner = mll.NelderMeadLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, learner_archive_filename='training_learner_archive', learner_archive_file_type=learner_archive_file_type, **self.remaining_kwargs) else: self.log.error( 'Unknown training type provided to Gaussian process controller:' + repr(training_type)) self.archive_dict.update({'training_type': self.training_type}) self._update_controller_with_learner_attributes() self.gp_learner = mll.GaussianProcessLearner( start_datetime=self.start_datetime, num_params=num_params, min_boundary=min_boundary, max_boundary=max_boundary, trust_region=trust_region, learner_archive_filename=learner_archive_filename, learner_archive_file_type=learner_archive_file_type, **self.remaining_kwargs) self.gp_learner_params_queue = self.gp_learner.params_out_queue self.gp_learner_costs_queue = self.gp_learner.costs_in_queue self.end_gp_learner = self.gp_learner.end_event self.new_params_event = self.gp_learner.new_params_event self.remaining_kwargs = self.gp_learner.remaining_kwargs self.generation_num = self.gp_learner.generation_num