def runModel(self, kind, conf):
self.setDatasets(conf)
# Create the experiment
exp = self.experiment
name = 'AL%d-Iter%d-%s' % (exp.experiment_id,
self.iteration.iteration_number, kind)
model_exp = ClassificationExperiment(exp.secuml_conf,
session=exp.session)
model_exp.initExperiment(exp.project,
exp.dataset,
experiment_name=name,
parent=exp.experiment_id)
model_exp.setConf(conf,
exp.features_filename,
annotations_id=exp.annotations_id)
model_exp.export()
# Build the model
model = conf.model_class(model_exp.conf, cv_monitoring=True)
model_run = RunClassifier(model, self.datasets, model_exp)
model_run.run()
self.models[kind] = model
# Execution time monitoring
time = model.training_execution_time + model.testing_execution_time
self.times[kind] = time
return model_exp
def createNaiveBayesConf(self):
exp = self.experiment
name = '-'.join([
'AL' + str(exp.experiment_id),
'Iter' + str(self.iteration.iteration_number), 'all', 'NaiveBayes'
])
naive_bayes_exp = ClassificationExperiment(exp.secuml_conf,
session=exp.session)
naive_bayes_exp.initExperiment(exp.project,
exp.dataset,
experiment_name=name,
parent=exp.experiment_id)
test_conf = UnlabeledLabeledConf(logger=exp.logger)
naive_bayes_conf = GaussianNaiveBayesConfiguration(
exp.conf.models_conf['multiclass'].n_jobs,
exp.conf.models_conf['multiclass'].num_folds,
False,
True,
test_conf,
logger=exp.logger)
naive_bayes_exp.setConf(naive_bayes_conf,
exp.features_filename,
annotations_id=exp.annotations_id)
naive_bayes_exp.export()
return naive_bayes_conf
def getMulticlassConf(self):
conf = self.rare_category_detection_conf.classification_conf
exp = self.experiment
name = '-'.join(['AL' + str(exp.experiment_id),
'Iter' + str(self.iteration.iteration_number),
self.label,
'analysis'])
self.multiclass_exp = ClassificationExperiment(exp.project, exp.dataset,
exp.session,
experiment_name=name,
parent=exp.experiment_id)
self.multiclass_exp.setConf(conf, exp.features_filename,
annotations_id=exp.annotations_id)
self.multiclass_exp.export()
return conf
def getNaiveBayesConf(self):
exp = self.experiment
name = '-'.join([
'AL' + str(exp.experiment_id),
'Iter' + str(self.iteration.iteration_number), 'all', 'NaiveBayes'
])
naive_bayes_exp = ClassificationExperiment(exp.project,
exp.dataset,
exp.session,
experiment_name=name,
parent=exp.experiment_id)
test_conf = UnlabeledLabeledConf()
naive_bayes_conf = GaussianNaiveBayesConfiguration(
exp.conf.num_folds, False, True, test_conf)
naive_bayes_exp.setConf(naive_bayes_conf,
exp.features_filename,
annotations_id=exp.annotations_id)
naive_bayes_exp.export()
return naive_bayes_conf
class RareCategoryDetectionAnnotationQueriesExp(
RareCategoryDetectionAnnotationQueries):
def __init__(self,
iteration,
label,
proba_min,
proba_max,
multiclass_exp=None,
input_checking=True):
RareCategoryDetectionAnnotationQueries.__init__(
self,
iteration,
label,
proba_min,
proba_max,
input_checking=input_checking)
self.multiclass_exp = multiclass_exp
self.experiment = iteration.experiment
def generateAnnotationQuery(self,
instance_id,
predicted_proba,
suggested_label,
suggested_family,
confidence=None):
return AnnotationQueryExp(instance_id,
predicted_proba,
suggested_label,
suggested_family,
confidence=confidence)
def getMulticlassConf(self):
conf = self.rare_category_detection_conf.classification_conf
exp = self.experiment
name = '-'.join([
'AL%d' % exp.experiment_id,
'Iter%d' % self.iteration.iteration_number, self.label, 'analysis'
])
self.multiclass_exp = ClassificationExperiment(exp.secuml_conf,
session=exp.session)
self.multiclass_exp.initExperiment(exp.project,
exp.dataset,
experiment_name=name,
parent=exp.experiment_id)
self.multiclass_exp.setConf(conf,
exp.features_filename,
annotations_id=exp.annotations_id)
self.multiclass_exp.export()
return conf
def createClusteringExperiment(self):
exp = self.experiment
conf = ClusteringConfiguration(self.categories.numCategories(),
logger=exp.logger)
name = '-'.join([
'AL%d' % exp.experiment_id,
'Iter%d' % self.iteration.iteration_number, self.label,
'clustering'
])
clustering_exp = ClusteringExperiment(exp.secuml_conf,
session=exp.session)
clustering_exp.initExperiment(exp.project,
exp.dataset,
experiment_name=name,
parent=exp.experiment_id)
clustering_exp.setConf(conf,
exp.features_filename,
annotations_id=exp.annotations_id)
clustering_exp.export()
return clustering_exp
def generateClusteringVisualization(self):
if self.families_analysis:
self.clustering_exp = self.createClusteringExperiment()
clustering = Clustering(self.categories.instances,
self.categories.assigned_categories)
clustering.generateClustering(None, None)
clustering.export(self.clustering_exp.getOutputDirectory())
else:
self.clustering_exp = None
def setCategories(self, all_instances, assigned_categories,
predicted_proba):
self.categories = CategoriesExp(self.multiclass_exp, self.iteration,
all_instances, assigned_categories,
predicted_proba, self.label,
self.multiclass_model.class_labels)