alibi.explainers.anchors.anchor_base

Constants

logger

logger: logging.Logger = <Logger alibi.explainers.anchors.anchor_base (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.

AnchorBaseBeam

Constructor

AnchorBaseBeam(self, samplers: List[Callable], **kwargs) -> None

Name	Type	Default	Description
samplers	List[Callable]		Objects that can be called with args (result, n_samples) tuple to draw samples.

Methods

anchor_beam

anchor_beam(delta: float = 0.05, epsilon: float = 0.1, desired_confidence: float = 1.0, beam_size: int = 1, epsilon_stop: float = 0.05, min_samples_start: int = 100, max_anchor_size: Optional[int] = None, stop_on_first: bool = False, batch_size: int = 100, coverage_samples: int = 10000, verbose: bool = False, verbose_every: int = 1, kwargs) -> dict

Name	Type	Default	Description
delta	float	0.05	Used to compute beta.
epsilon	float	0.1	Precision bound tolerance for convergence.
desired_confidence	float	1.0	Desired level of precision (tau in paper <https://homes.cs.washington.edu/~marcotcr/aaai18.pdf>_).
beam_size	int	1	Beam width.
epsilon_stop	float	0.05	Confidence bound margin around desired precision.
min_samples_start	int	100	Min number of initial samples.
max_anchor_size	Optional[int]	None	Max number of features in result.
stop_on_first	bool	False	Stop on first valid result found.
batch_size	int	100	Number of samples used for an arm evaluation.
coverage_samples	int	10000	Number of samples from which to build a coverage set.
verbose	bool	False	Whether to print intermediate LUCB & anchor selection output.
verbose_every	int	1	Print intermediate output every verbose_every steps.

Returns

• Type: dict

compute_beta

compute_beta(n_features: int, t: int, delta: float) -> float

Name	Type	Default	Description
n_features	int		Number of candidate anchors.
t	int		Iteration number.
delta	float		Confidence budget, candidate anchors have close to optimal precisions with prob. 1 - delta.

Returns

• Type: float

dlow_bernoulli

dlow_bernoulli(p: numpy.ndarray, level: numpy.ndarray, n_iter: int = 17) -> numpy.ndarray

Name	Type	Default	Description
p	numpy.ndarray		Precision of candidate anchors.
level	numpy.ndarray		beta / nb of samples for each result.
n_iter	int	17	Number of iterations during lower bound update.

Returns

• Type: numpy.ndarray

draw_samples

draw_samples(anchors: list, batch_size: int) -> Tuple[tuple, tuple]

Name	Type	Default	Description
anchors	list		Anchors on which samples are conditioned.
batch_size	int		The number of samples drawn for each result.

Returns

• Type: Tuple[tuple, tuple]

dup_bernoulli

dup_bernoulli(p: numpy.ndarray, level: numpy.ndarray, n_iter: int = 17) -> numpy.ndarray

Name	Type	Default	Description
p	numpy.ndarray		Precision of candidate anchors.
level	numpy.ndarray		beta / nb of samples for each result.
n_iter	int	17	Number of iterations during lower bound update.

Returns

• Type: numpy.ndarray

get_anchor_metadata

get_anchor_metadata(features: tuple, success, batch_size: int = 100) -> dict

Name	Type	Default	Description
features	tuple		Sorted indices of features in result.
success			Indicates whether an anchor satisfying precision threshold was met or not.
batch_size	int	100	Number of samples among which positive and negative examples for partial anchors are selected if partial anchors have not already been explicitly sampled.

Returns

• Type: dict

get_init_stats

get_init_stats(anchors: list, coverages = False) -> dict

Name	Type	Default	Description
anchors	list		Candidate anchors.
coverages		False	If True, the statistics returned contain the coverage of the specified anchors.

Returns

• Type: dict

kllucb

kllucb(anchors: list, init_stats: dict, epsilon: float, delta: float, batch_size: int, top_n: int, verbose: bool = False, verbose_every: int = 1) -> numpy.ndarray

Name	Type	Default	Description
anchors	list		A list of anchors from which two critical anchors are selected (see Kaufmann and Kalyanakrishnan, 2013).
init_stats	dict		Dictionary with lists containing nb of samples used and where sample predictions equal the desired label.
epsilon	float		Precision bound tolerance for convergence.
delta	float		Used to compute beta.
batch_size	int		Number of samples.
top_n	int		Min of beam width size or number of candidate anchors.
verbose	bool	False	Whether to print intermediate output.
verbose_every	int	1	Whether to print intermediate output every verbose_every steps.

Returns

• Type: numpy.ndarray

propose_anchors

propose_anchors(previous_best: list) -> list

Name	Type	Default	Description
previous_best	list		List with tuples of result candidates.

Returns

• Type: list

select_critical_arms

select_critical_arms(means: numpy.ndarray, ub: numpy.ndarray, lb: numpy.ndarray, n_samples: numpy.ndarray, delta: float, top_n: int, t: int)

Name	Type	Default	Description
means	numpy.ndarray		Empirical mean result precisions.
ub	numpy.ndarray		Upper bound on result precisions.
lb	numpy.ndarray		Lower bound on result precisions.
n_samples	numpy.ndarray		The number of samples drawn for each candidate result.
delta	float		Confidence budget, candidate anchors have close to optimal precisions with prob. 1 - delta.
top_n	int		Number of arms to be selected.
t	int		Iteration number.

to_sample

to_sample(means: numpy.ndarray, ubs: numpy.ndarray, lbs: numpy.ndarray, desired_confidence: float, epsilon_stop: float)

Name	Type	Default	Description
means	numpy.ndarray		Mean precisions (each element represents a different result).
ubs	numpy.ndarray		Precisions' upper bounds (each element represents a different result).
lbs	numpy.ndarray		Precisions' lower bounds (each element represents a different result).
desired_confidence	float		Desired level of confidence for precision estimation.
epsilon_stop	float		Tolerance around desired precision.

update_state

update_state(covered_true: numpy.ndarray, covered_false: numpy.ndarray, labels: numpy.ndarray, samples: Tuple[numpy.ndarray, float], anchor: tuple) -> Tuple[int, int]

Name	Type	Default	Description
covered_true	numpy.ndarray		Examples where the result applies and the prediction is the same as on the instance to be explained.
covered_false	numpy.ndarray		Examples where the result applies and the prediction is the different to the instance to be explained.
labels	numpy.ndarray		An array indicating whether the prediction on the sample matches the label of the instance to be explained.
samples	Tuple[numpy.ndarray, float]		A tuple containing discretized data, coverage and the result sampled.
anchor	tuple		The result to be updated.

Returns

• Type: Tuple[int, int]