TY - GEN
T1 - A formal approach to explainability
AU - Wolf, Lior
AU - Galanti, Tomer
AU - Hazan, Tamir
N1 - Publisher Copyright: © 2019 Copyright is held by the owner/author(s).
PY - 2019/1/27
Y1 - 2019/1/27
N2 - We regard explanations as a blending of the input sample and the model's output and offer a few definitions that capture various desired properties of the function that generates these explanations. We study the links between these properties and between explanation-generating functions and intermediate representations of learned models and are able to show, for example, that if the activations of a given layer are consistent with an explanation, then so do all other subsequent layers. In addition, we study the intersection and union of explanations as a way to construct new explanations.
AB - We regard explanations as a blending of the input sample and the model's output and offer a few definitions that capture various desired properties of the function that generates these explanations. We study the links between these properties and between explanation-generating functions and intermediate representations of learned models and are able to show, for example, that if the activations of a given layer are consistent with an explanation, then so do all other subsequent layers. In addition, we study the intersection and union of explanations as a way to construct new explanations.
UR - http://www.scopus.com/inward/record.url?scp=85070596198&partnerID=8YFLogxK
U2 - https://doi.org/10.1145/3306618.3314260
DO - https://doi.org/10.1145/3306618.3314260
M3 - منشور من مؤتمر
T3 - AIES 2019 - Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society
SP - 255
EP - 261
BT - AIES 2019 - Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society
T2 - 2nd AAAI/ACM Conference on AI, Ethics, and Society, AIES 2019
Y2 - 27 January 2019 through 28 January 2019
ER -