Custom Types

Custom enums of supported functionality and types of for the ecgan library.

class ecgan.utils.custom_types.AnomalyDetectionStrategies(value)

Bases: enum.Enum

Implemented strategies for anomaly detection.

class ecgan.utils.custom_types.Transformation(value)

Bases: enum.Enum

Implemented transformations which can be set before passing data to a module.

class ecgan.utils.custom_types.SplitMethods(value)

Bases: enum.Enum

Available split methods.

class ecgan.utils.custom_types.SampleDataset(value)

Bases: enum.Enum

Available datasets a DatasetSampler can sample from.

class ecgan.utils.custom_types.InverseMappingType(value)

Bases: enum.Enum

Available inverse mappings.

class ecgan.utils.custom_types.ReconstructionType(value)

Bases: enum.Enum

Different ways to reconstruct data using an already trained GAN generator.

class ecgan.utils.custom_types.DiscriminationStrategy(value)

Bases: enum.Enum

Different ways to discriminate data using an already trained GAN discriminator.

class ecgan.utils.custom_types.SimilarityMeasures(value)

Bases: enum.Enum

Methods to calculate the similarity between two series.

class ecgan.utils.custom_types.Optimizers(value)

Bases: enum.Enum

Supported optimizers.

class ecgan.utils.custom_types.MetricOptimization(value)

Bases: enum.Enum

Supported methods to optimize weighted errors regarding their weighting for a total anomaly score.

class ecgan.utils.custom_types.Losses(value)

Bases: enum.Enum

Supported losses.

class ecgan.utils.custom_types.LabelingStrategy(value)

Bases: enum.Enum

Determine how the points shall be labeled.

WARNING: The strategy chosen has implications on the output format. POINTWISE returns a label for each datapoint in each series. ACCUMULATE_UNIVARIATE returns a label for each univariate series. ACCUMULATE_MULTIVARIATE returns a label for each multivariate series.

class ecgan.utils.custom_types.SamplingAlgorithm(value)

Bases: enum.Enum

Different down- or upsampling algorithms which can be used during preprocessing.

class ecgan.utils.custom_types.MetricType(value)

Bases: enum.Enum

Supported evaluation metrics.

class ecgan.utils.custom_types.WeightInitialization(value)

Bases: enum.Enum

Different strategies to inititalize weights in neural networks.

class ecgan.utils.custom_types.InputNormalization(value)

Bases: enum.Enum

Supported normalization layers.

class ecgan.utils.custom_types.LatentDistribution(value)

Bases: enum.Enum

Supported latent distributions which can be set via config.

class ecgan.utils.custom_types.TrackerType(value)

Bases: enum.Enum

Supported trackers that can be set via config.

class ecgan.utils.custom_types.PlotterType(value)

Bases: enum.Enum

Different types of plotters.

class ecgan.utils.custom_types.SklearnAveragingOptions(value)

Bases: enum.Enum

Additional Sklearn type stubs for the available F-score averaging options.

More information on the respective effects: https://scikit-learn.org/stable/modules/generated/sklearn.metrics.precision_recall_fscore_support.html

class ecgan.utils.custom_types.SklearnSVMKernels(value)

Bases: enum.Enum

SVM kernels supported by ECGAN from sklearn.

More informations can be found in the official [docs] (https://scikit-learn.org/stable/modules/generated/sklearn.svm.SVC.html)

class ecgan.utils.custom_types.SupportedModules(value)

Bases: enum.Enum

Modules supported by the framework.