nessie.detectors.classification_entropy

Module Contents

Classes

ClassificationEntropy

Given a distribution over labels for each instance in form of a (num_instances, num_labels)

class nessie.detectors.classification_entropy.ClassificationEntropy

Bases: nessie.detectors.error_detector.ModelBasedDetector

Given a distribution over labels for each instance in form of a (num_instances, num_labels) numpy array, the resulting score for each instance is the entropy of each instances label distribution. If the entropy is larger, then this means more uncertainty and higher chance of being an annotation error.

Active Learning Book Synthesis Lectures on Artificial Intelligence and Machine Learning Morgan & Claypool Publishers, June 2012 Section 2.3

See also https://modal-python.readthedocs.io/en/latest/content/query_strategies/uncertainty_sampling.html

error_detector_kind(self)

score(self, probabilities: numpy.typing.NDArray[float], **kwargs) → numpy.typing.NDArray[float]

Scores the input according to their class distribution entropy.

Parameters

probabilities – a (num_instances, num_classes) numpy array obtained from a machine learning model

Returns

a (num_instances,) numpy array containing the resulting scores

Return type

scores

uses_probabilities(self) → bool