The explanation game: a formal framework for interpretable machine learning.
We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation(s) for a given algorithmic prediction. Throug...
| Publicado en: | Synthese Vol. 198; no. 10; pp. 9211 - 9243 |
|---|---|
| Autores principales: | , |
| Formato: | Artículo |
| Publicado: |
Springer Nature
Oct2021
|
| Materias: | |
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=hlh&AN=152559352&site=ehost-live header: @attributes: shortDbName: hlh uiTerm: 152559352 longDbName: Humanities International Complete uiTag: AN controlInfo: bkinfo: jinfo: jid: 00397857 4LI jtl: Synthese issn: 00397857 maglogo: N pubinfo: dt: Oct2021 vid: 198 iid: 10 pid: 237 pub: Springer Nature artinfo: ui: 152559352 10.1007/s11229-020-02629-9 ppf: 9211 ppct: 32 formats: fmt: @attributes: type: P size: 618KB tig: atl: The explanation game: a formal framework for interpretable machine learning. aug: au: Watson, David S. Floridi, Luciano affil: Oxford Internet Institute, University of Oxford, 41 Saint Giles, OX1 3LW, Oxford, UK The Alan Turing Institute, British Library, 96 Euston Road, Kings Cross, NW1 2DB, London, UK su: Machine learning Decision theory Statistical learning Polynomial time algorithms Explanation sug: subj: Machine learning Decision theory Statistical learning Polynomial time algorithms Explanation keyword: Algorithmic explainability Explanation game Interpretable machine learning Pareto frontier Relevance ab: We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. pubtype: Academic Journal doctype: Article src: R language: English refInfo: copyright: @attributes: flag: Y custom: Synthese is a copyright of Springer, 2021. All Rights Reserved. item: Synthese holder: Springer Nature dt: @attributes: year: 2021 holdings: @attributes: islocal: N |
|---|