Understanding Post-hoc Explainers: The Case of Anchors
| Autor: | Lopardo, Gianluigi, Precioso, Frederic, Garreau, Damien |
|---|---|
| Přispěvatelé: | Modèles et algorithmes pour l’intelligence artificielle (MAASAI), Inria Sophia Antipolis - Méditerranée (CRISAM), Institut National de Recherche en Informatique et en Automatique (Inria)-Institut National de Recherche en Informatique et en Automatique (Inria)-Université Nice Sophia Antipolis (1965 - 2019) (UNS), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Laboratoire Jean Alexandre Dieudonné (LJAD), Université Nice Sophia Antipolis (1965 - 2019) (UNS), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)-Scalable and Pervasive softwARe and Knowledge Systems (Laboratoire I3S - SPARKS), Laboratoire d'Informatique, Signaux, et Systèmes de Sophia Antipolis (I3S), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)-Université Nice Sophia Antipolis (1965 - 2019) (UNS), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)-Laboratoire d'Informatique, Signaux, et Systèmes de Sophia Antipolis (I3S), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA)-Centre National de la Recherche Scientifique (CNRS), Laboratoire Jean Alexandre Dieudonné (LJAD), COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-COMUE Université Côte d'Azur (2015-2019) (COMUE UCA)-Centre National de la Recherche Scientifique (CNRS)-Université Côte d'Azur (UCA), Université Côte d'Azur (UCA), ANR-21-CE23-0005,NIM-ML,Méthodes d'explicabilité de nouvelle génération(2021), European Project: H2020-951911,H2020-EU.2.1.1. - INDUSTRIAL LEADERSHIP - Leadership in enabling and industrial technologies - Information and Communication Technologies (ICT),AI4Media(2020) |
| Jazyk: | angličtina |
| Rok vydání: | 2023 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Machine Learning Computer Science - Computation and Language Apprentissage statistique Computer Science - Artificial Intelligence Machine Learning (stat.ML) Classification supervisée et non supervisée Statistical learning Machine Learning (cs.LG) Artificial Intelligence (cs.AI) Interprétabilité [INFO.INFO-LG]Computer Science [cs]/Machine Learning [cs.LG] Statistics - Machine Learning Supervised and unsupervised classification Interpretability Computation and Language (cs.CL) Traitement du Langage Naturel Natural Language Processing |
| Popis: | In many scenarios, the interpretability of machine learning models is a highly required but difficult task. To explain the individual predictions of such models, local model-agnostic approaches have been proposed. However, the process generating the explanations can be, for a user, as mysterious as the prediction to be explained. Furthermore, interpretability methods frequently lack theoretical guarantees, and their behavior on simple models is frequently unknown. While it is difficult, if not impossible, to ensure that an explainer behaves as expected on a cutting-edge model, we can at least ensure that everything works on simple, already interpretable models. In this paper, we present a theoretical analysis of Anchors (Ribeiro et al., 2018): a popular rule-based interpretability method that highlights a small set of words to explain a text classifier's decision. After formalizing its algorithm and providing useful insights, we demonstrate mathematically that Anchors produces meaningful results when used with linear text classifiers on top of a TF-IDF vectorization. We believe that our analysis framework can aid in the development of new explainability methods based on solid theoretical foundations. arXiv admin note: substantial text overlap with arXiv:2205.13789 |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|