Explaining Naive Bayes and Other Linear Classifiers with Polynomial Time and Delay - Université Toulouse III - Paul Sabatier - Toulouse INP Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

Explaining Naive Bayes and Other Linear Classifiers with Polynomial Time and Delay

Résumé

Recent work proposed the computation of so-called PI-explanations of Naive Bayes Classifiers (NBCs) [35]. PI-explanations are subset-minimal sets of feature-value pairs that are sufficient for the prediction, and have been computed with stateof-the-art exact algorithms that are worst-case exponential in time and space. In contrast, we show that the computation of one PI-explanation for an NBC can be achieved in log-linear time, and that the same result also applies to the more general class of linear classifiers. Furthermore, we show that the enumeration of PI-explanations can be obtained with polynomial delay. Experimental results demonstrate the performance gains of the new algorithms when compared with earlier work. The experimental results also investigate ways to measure the quality of heuristic explanations.
Fichier principal
Vignette du fichier
NeurIPS-2020-explaining-naive-bayes-and-other-linear-classifiers-with-polynomial-time-and-delay-Paper.pdf (547.83 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03311391 , version 1 (31-07-2021)

Licence

Paternité

Identifiants

  • HAL Id : hal-03311391 , version 1

Citer

Joao Marques-Silva, Thomas Gerspacher, Martin Cooper, Alexey Ignatiev, Nina Narodytska. Explaining Naive Bayes and Other Linear Classifiers with Polynomial Time and Delay. 34th Conference on Neural Information Processing Systems (NeurIPS 2020), 2020, Vancouver (virtual), Canada. pp.1-11. ⟨hal-03311391⟩
36 Consultations
37 Téléchargements

Partager

Gmail Facebook X LinkedIn More