alibi_detect.od.llr

Constants

logger

logger: logging.Logger = <Logger alibi_detect.od.llr (WARNING)>

Instances of the Logger class represent a single logging channel. A "logging channel" indicates an area of an application. Exactly how an "area" is defined is up to the application developer. Since an application can have any number of areas, logging channels are identified by a unique string. Application areas can be nested (e.g. an area of "input processing" might include sub-areas "read CSV files", "read XLS files" and "read Gnumeric files"). To cater for this natural nesting, channel names are organized into a namespace hierarchy where levels are separated by periods, much like the Java or Python package namespace. So in the instance given above, channel names might be "input" for the upper level, and "input.csv", "input.xls" and "input.gnu" for the sub-levels. There is no arbitrary limit to the depth of nesting.

LLR

Inherits from: BaseDetector, FitMixin, ThresholdMixin, ABC

Constructor

LLR(self, threshold: float = None, model: Union[keras.src.models.model.Model, tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN] = None, model_background: Union[keras.src.models.model.Model, tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN] = None, log_prob: Callable = None, sequential: bool = False, data_type: str = None) -> None

Name	Type	Default	Description
threshold	Optional[float]	None	Threshold used for the likelihood ratio (LLR) to determine outliers.
model	Union[keras.src.models.model.Model, tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN, None]	None	Generative model, defaults to PixelCNN.
model_background	Union[keras.src.models.model.Model, tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN, None]	None	Optional model for the background. Only needed if it is different from model.
log_prob	Optional[Callable]	None	Function used to evaluate log probabilities under the model if the model does not have a log_prob function.
sequential	bool	False	Whether the data is sequential. Used to create targets during training.
data_type	Optional[str]	None	Optionally specify the data type (tabular, image or time-series). Added to metadata.

Methods

feature_score

feature_score(X: numpy.ndarray, batch_size: int = 10000000000) -> numpy.ndarray

Feature-level negative likelihood ratios.

Name	Type	Default	Description
X	numpy.ndarray
batch_size	int	10000000000

Returns

• Type: numpy.ndarray

fit

fit(X: numpy.ndarray, mutate_fn: Callable = <function mutate_categorical at 0x16ddcdaf0>, mutate_fn_kwargs: dict = {'rate': 0.2, 'seed': 0, 'feature_range': (0, 255)}, mutate_batch_size: int = 10000000000, loss_fn: .tensorflow.keras.losses = None, loss_fn_kwargs: dict = None, optimizer: Union[ForwardRef('tf.keras.optimizers.Optimizer'), ForwardRef('tf.keras.optimizers.legacy.Optimizer'), type[ForwardRef('tf.keras.optimizers.Optimizer')], type[ForwardRef('tf.keras.optimizers.legacy.Optimizer')]] = <class 'keras.src.optimizers.adam.Adam'>, epochs: int = 20, batch_size: int = 64, verbose: bool = True, log_metric: Tuple[str, ForwardRef('tf.keras.metrics')] = None, callbacks: .tensorflow.keras.callbacks = None) -> None

Train semantic and background generative models.

Name	Type	Default	Description
X	numpy.ndarray		Training batch.
mutate_fn	Callable	<function mutate_categorical at 0x16ddcdaf0>	Mutation function used to generate the background dataset.
mutate_fn_kwargs	dict	{'rate': 0.2, 'seed': 0, 'feature_range': (0, 255)}	Kwargs for the mutation function used to generate the background dataset. Default values set for an image dataset.
mutate_batch_size	int	10000000000	Batch size used to generate the mutations for the background dataset.
loss_fn	.tensorflow.keras.losses	None	Loss function used for training.
loss_fn_kwargs	dict	None	Kwargs for loss function.
optimizer	Union[ForwardRef('tf.keras.optimizers.Optimizer'), ForwardRef('tf.keras.optimizers.legacy.Optimizer'), type[ForwardRef('tf.keras.optimizers.Optimizer')], type[ForwardRef('tf.keras.optimizers.legacy.Optimizer')]]	<class 'keras.src.optimizers.adam.Adam'>	Optimizer used for training.
epochs	int	20	Number of training epochs.
batch_size	int	64	Batch size used for training.
verbose	bool	True	Whether to print training progress.
log_metric	Tuple[str, ForwardRef('tf.keras.metrics')]	None	Additional metrics whose progress will be displayed if verbose equals True.
callbacks	.tensorflow.keras.callbacks	None	Callbacks used during training.

Returns

• Type: None

infer_threshold

infer_threshold(X: numpy.ndarray, outlier_type: str = 'instance', threshold_perc: float = 95.0, batch_size: int = 10000000000) -> None

Update LLR threshold by a value inferred from the percentage of instances

considered to be outliers in a sample of the dataset.

Name	Type	Default	Description
X	numpy.ndarray		Batch of instances.
outlier_type	str	'instance'	Predict outliers at the 'feature' or 'instance' level.
threshold_perc	float	95.0	Percentage of sorted feature level outlier scores used to predict instance level outlier.
batch_size	int	10000000000	Batch size for the generative model evaluations.

Returns

• Type: None

instance_score

instance_score(X: numpy.ndarray, batch_size: int = 10000000000) -> numpy.ndarray

Instance-level negative likelihood ratios.

Name	Type	Default	Description
X	numpy.ndarray
batch_size	int	10000000000

Returns

• Type: numpy.ndarray

llr

llr(X: numpy.ndarray, return_per_feature: bool, batch_size: int = 10000000000) -> numpy.ndarray

Compute likelihood ratios.

Name	Type	Default	Description
X	numpy.ndarray		Batch of instances.
return_per_feature	bool		Return likelihood ratio per feature.
batch_size	int	10000000000	Batch size for the generative model evaluations.

Returns

• Type: numpy.ndarray

logp

logp(dist, X: numpy.ndarray, return_per_feature: bool = False, batch_size: int = 10000000000) -> numpy.ndarray

Compute log probability of a batch of instances under the generative model.

Name	Type	Default	Description
dist			Distribution of the model.
X	numpy.ndarray		Batch of instances.
return_per_feature	bool	False	Return log probability per feature.
batch_size	int	10000000000	Batch size for the generative model evaluations.

Returns

• Type: numpy.ndarray

logp_alt

logp_alt(model: keras.src.models.model.Model, X: numpy.ndarray, return_per_feature: bool = False, batch_size: int = 10000000000) -> numpy.ndarray

Compute log probability of a batch of instances using the log_prob function

defined by the user.

Name	Type	Default	Description
model	keras.src.models.model.Model		Trained model.
X	numpy.ndarray		Batch of instances.
return_per_feature	bool	False	Return log probability per feature.
batch_size	int	10000000000	Batch size for the generative model evaluations.

Returns

• Type: numpy.ndarray

predict

predict(X: numpy.ndarray, outlier_type: str = 'instance', batch_size: int = 10000000000, return_feature_score: bool = True, return_instance_score: bool = True) -> Dict[Dict[str, str], Dict[numpy.ndarray, numpy.ndarray]]

Predict whether instances are outliers or not.

Name	Type	Default	Description
X	numpy.ndarray		Batch of instances.
outlier_type	str	'instance'	Predict outliers at the 'feature' or 'instance' level.
batch_size	int	10000000000	Batch size used when making predictions with the generative model.
return_feature_score	bool	True	Whether to return feature level outlier scores.
return_instance_score	bool	True	Whether to return instance level outlier scores.

Returns

• Type: Dict[Dict[str, str], Dict[numpy.ndarray, numpy.ndarray]]

score

score(X: numpy.ndarray, batch_size: int = 10000000000) -> Tuple[numpy.ndarray, numpy.ndarray]

Feature-level and instance-level outlier scores.

The scores are equal to the negative likelihood ratios.

Name	Type	Default	Description
X	numpy.ndarray
batch_size	int	10000000000

Returns

• Type: Tuple[numpy.ndarray, numpy.ndarray]

Functions

build_model

build_model(dist: Union[tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN], input_shape: Optional[tuple] = None, filepath: Optional[str] = None) -> Tuple[keras.src.models.model.Model, Union[tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN]]

Create tf.keras.Model from TF distribution.

Name	Type	Default	Description
dist	Union[tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN]		TensorFlow distribution.
input_shape	Optional[tuple]	None	Input shape of the model.
filepath	Optional[str]	None	File to load model weights from.

Returns

• Type: Tuple[keras.src.models.model.Model, Union[tensorflow_probability.python.distributions.distribution.Distribution, alibi_detect.models.tensorflow.pixelcnn.PixelCNN]]