Privacy-Preserving Contrastive Explanations with Local Foil Trees

Thijs Veugen; Bart Kamphorst; Michiel Marcus

doi:10.3390/cryptography6040054

Privacy-Preserving Contrastive Explanations with Local Foil Trees

Thijs Veugen^*
, Bart Kamphorst
, Michiel Marcus

^*Corresponding author for this work

Semantics, Cybersecurity & Services

Research output: Contribution to journal › Article › Academic › peer-review

2 Citations (Scopus)

82 Downloads (Pure)

Abstract

We present the first algorithm that combines privacy-preserving technologies and state-of-the-art explainable AI to enable privacy-friendly explanations of black-box AI models. We provide a secure algorithm for contrastive explanations of black-box machine learning models that securely trains and uses local foil trees. Our work shows that the quality of these explanations can be upheld whilst ensuring the privacy of both the training data and the model itself.

Original language	English
Article number	54
Journal	Cryptography
Volume	6
Issue number	4
DOIs	https://doi.org/10.3390/cryptography6040054
Publication status	Published - Oct 2022

Keywords

Explainable AI
Secure multi-party computation
Decision tree
Foil tree

Access to Document

10.3390/cryptography6040054Licence: CC BY

ArticleFinal published version, 395 KBLicence: CC BY

Cite this

@article{bbf154d116424ee08e68398f74be65e4,

title = "Privacy-Preserving Contrastive Explanations with Local Foil Trees",

abstract = "We present the first algorithm that combines privacy-preserving technologies and state-of-the-art explainable AI to enable privacy-friendly explanations of black-box AI models. We provide a secure algorithm for contrastive explanations of black-box machine learning models that securely trains and uses local foil trees. Our work shows that the quality of these explanations can be upheld whilst ensuring the privacy of both the training data and the model itself.",

keywords = "Explainable AI, Secure multi-party computation, Decision tree, Foil tree",

author = "Thijs Veugen and Bart Kamphorst and Michiel Marcus",

note = "Publisher Copyright: {\textcopyright} 2022 by the authors.",

year = "2022",

month = oct,

doi = "10.3390/cryptography6040054",

language = "English",

volume = "6",

journal = "Cryptography",

issn = "2410-387X",

publisher = "MDPI",

number = "4",

}

TY - JOUR

T1 - Privacy-Preserving Contrastive Explanations with Local Foil Trees

AU - Veugen, Thijs

AU - Kamphorst, Bart

AU - Marcus, Michiel

PY - 2022/10

Y1 - 2022/10

N2 - We present the first algorithm that combines privacy-preserving technologies and state-of-the-art explainable AI to enable privacy-friendly explanations of black-box AI models. We provide a secure algorithm for contrastive explanations of black-box machine learning models that securely trains and uses local foil trees. Our work shows that the quality of these explanations can be upheld whilst ensuring the privacy of both the training data and the model itself.

AB - We present the first algorithm that combines privacy-preserving technologies and state-of-the-art explainable AI to enable privacy-friendly explanations of black-box AI models. We provide a secure algorithm for contrastive explanations of black-box machine learning models that securely trains and uses local foil trees. Our work shows that the quality of these explanations can be upheld whilst ensuring the privacy of both the training data and the model itself.

KW - Explainable AI

KW - Secure multi-party computation

KW - Decision tree

KW - Foil tree

UR - https://www.scopus.com/pages/publications/85144733621

U2 - 10.3390/cryptography6040054

DO - 10.3390/cryptography6040054

M3 - Article

SN - 2410-387X

VL - 6

JO - Cryptography

JF - Cryptography

IS - 4

M1 - 54

ER -

Privacy-Preserving Contrastive Explanations with Local Foil Trees

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Privacy-Preserving Contrastive Explanations with Local Foil Trees

Cite this