Local Explanations via Necessity and Sufficiency: Unifying Theory and Practice

David S. Watson; Limor Gultchin; Ankur Taly; Luciano Floridi

Local Explanations via Necessity and Sufficiency: Unifying Theory and Practice

David S. Watson, Limor Gultchin, Ankur Taly, Luciano Floridi

Informatics

Research output: Contribution to conference types › Paper › peer-review

33 Citations (Scopus)

Abstract

Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks.

Original language	English
Pages	1382-1392
Number of pages	11
Publication status	Published - 2021
Event	37th Conference on Uncertainty in Artificial Intelligence, UAI 2021 - Virtual, Online Duration: 27 Jul 2021 → 30 Jul 2021

Conference

Conference	37th Conference on Uncertainty in Artificial Intelligence, UAI 2021
City	Virtual, Online
Period	27/07/2021 → 30/07/2021

Access to Document

https://proceedings.mlr.press/v161/watson21a.html

Cite this

@conference{f7e64ca28e074123abb91c36cd7285a4,

title = "Local Explanations via Necessity and Sufficiency: Unifying Theory and Practice",

abstract = "Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks.",

author = "Watson, {David S.} and Limor Gultchin and Ankur Taly and Luciano Floridi",

note = "Funding Information: DSW was supported by ONR grant N62909-19-1-2096. Publisher Copyright: {\textcopyright} 2021 37th Conference on Uncertainty in Artificial Intelligence, UAI 2021. All Rights Reserved.; 37th Conference on Uncertainty in Artificial Intelligence, UAI 2021 ; Conference date: 27-07-2021 Through 30-07-2021",

year = "2021",

language = "English",

pages = "1382--1392",

}

TY - CONF

T1 - Local Explanations via Necessity and Sufficiency

T2 - 37th Conference on Uncertainty in Artificial Intelligence, UAI 2021

AU - Watson, David S.

AU - Gultchin, Limor

AU - Taly, Ankur

AU - Floridi, Luciano

PY - 2021

Y1 - 2021

N2 - Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks.

AB - Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence (XAI), a fast-growing research area that is so far lacking in firm theoretical foundations. Building on work in logic, probability, and causality, we establish the central role of necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We provide a sound and complete algorithm for computing explanatory factors with respect to a given context, and demonstrate its flexibility and competitive performance against state of the art alternatives on various tasks.

UR - http://www.scopus.com/inward/record.url?scp=85124311740&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85124311740

SP - 1382

EP - 1392

Y2 - 27 July 2021 through 30 July 2021

ER -

Local Explanations via Necessity and Sufficiency: Unifying Theory and Practice

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this