Stochastic Integrated Explanations for Vision Models

Oren Barkan; Yehonatan Elisha; Jonathan Weill; Yuval Asher; Amit Eshel; Noam Koenigstein

doi:10.1109/ICDM58522.2023.00104

Stochastic Integrated Explanations for Vision Models

Oren Barkan, Yehonatan Elisha, Jonathan Weill, Yuval Asher, Amit Eshel, Noam Koenigstein

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

We introduce Stochastic Integrated Explanations (SIX) - a general method for explaining predictions made by vision models. SIX employs stochastic integration on the internal representations across different network layers, producing explanation maps at various scales. The primary innovation of SIX is the introduction of randomness to the integration process by modeling the baseline representation as a random tensor. Through iterative sampling from the baseline distribution, SIX generates a diverse set of explanation maps, allowing the selection of the best-performing map based on a specific metric of interest. Extensive evaluations on various model architectures showcase the superior performance of SIX compared to state-of-the-art explanation methods, affirming its effectiveness across multiple metrics. Our code is available at: https://github.com/six-icdm/six

Original language	English
Title of host publication	Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023
Editors	Guihai Chen, Latifur Khan, Xiaofeng Gao, Meikang Qiu, Witold Pedrycz, Xindong Wu
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	938-943
Number of pages	6
ISBN (Electronic)	9798350307887
DOIs	https://doi.org/10.1109/ICDM58522.2023.00104
State	Published - 2023
Externally published	Yes
Event	23rd IEEE International Conference on Data Mining, ICDM 2023 - Shanghai, China Duration: 1 Dec 2023 → 4 Dec 2023

Publication series

Name	Proceedings - IEEE International Conference on Data Mining, ICDM
ISSN (Print)	1550-4786

Conference

Conference	23rd IEEE International Conference on Data Mining, ICDM 2023
Country/Territory	China
City	Shanghai
Period	1/12/23 → 4/12/23

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Keywords

Computer Vision
Deep Learning
Explainable AI

Access to Document

10.1109/ICDM58522.2023.00104

Cite this

Barkan, O., Elisha, Y., Weill, J., Asher, Y., Eshel, A., & Koenigstein, N. (2023). Stochastic Integrated Explanations for Vision Models. In G. Chen, L. Khan, X. Gao, M. Qiu, W. Pedrycz, & X. Wu (Eds.), Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023 (pp. 938-943). (Proceedings - IEEE International Conference on Data Mining, ICDM). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICDM58522.2023.00104

Barkan, Oren ; Elisha, Yehonatan ; Weill, Jonathan et al. / Stochastic Integrated Explanations for Vision Models. Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023. editor / Guihai Chen ; Latifur Khan ; Xiaofeng Gao ; Meikang Qiu ; Witold Pedrycz ; Xindong Wu. Institute of Electrical and Electronics Engineers Inc., 2023. pp. 938-943 (Proceedings - IEEE International Conference on Data Mining, ICDM).

@inproceedings{b7a8f49acddd45a2a97870ee0f8796d0,

title = "Stochastic Integrated Explanations for Vision Models",

abstract = "We introduce Stochastic Integrated Explanations (SIX) - a general method for explaining predictions made by vision models. SIX employs stochastic integration on the internal representations across different network layers, producing explanation maps at various scales. The primary innovation of SIX is the introduction of randomness to the integration process by modeling the baseline representation as a random tensor. Through iterative sampling from the baseline distribution, SIX generates a diverse set of explanation maps, allowing the selection of the best-performing map based on a specific metric of interest. Extensive evaluations on various model architectures showcase the superior performance of SIX compared to state-of-the-art explanation methods, affirming its effectiveness across multiple metrics. Our code is available at: https://github.com/six-icdm/six",

keywords = "Computer Vision, Deep Learning, Explainable AI",

author = "Oren Barkan and Yehonatan Elisha and Jonathan Weill and Yuval Asher and Amit Eshel and Noam Koenigstein",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 23rd IEEE International Conference on Data Mining, ICDM 2023 ; Conference date: 01-12-2023 Through 04-12-2023",

year = "2023",

doi = "10.1109/ICDM58522.2023.00104",

language = "אנגלית",

series = "Proceedings - IEEE International Conference on Data Mining, ICDM",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "938--943",

editor = "Guihai Chen and Latifur Khan and Xiaofeng Gao and Meikang Qiu and Witold Pedrycz and Xindong Wu",

booktitle = "Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023",

address = "ארצות הברית",

}

Barkan, O, Elisha, Y, Weill, J, Asher, Y, Eshel, A & Koenigstein, N 2023, Stochastic Integrated Explanations for Vision Models. in G Chen, L Khan, X Gao, M Qiu, W Pedrycz & X Wu (eds), Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023. Proceedings - IEEE International Conference on Data Mining, ICDM, Institute of Electrical and Electronics Engineers Inc., pp. 938-943, 23rd IEEE International Conference on Data Mining, ICDM 2023, Shanghai, China, 1/12/23. https://doi.org/10.1109/ICDM58522.2023.00104

Stochastic Integrated Explanations for Vision Models. / Barkan, Oren; Elisha, Yehonatan; Weill, Jonathan et al.
Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023. ed. / Guihai Chen; Latifur Khan; Xiaofeng Gao; Meikang Qiu; Witold Pedrycz; Xindong Wu. Institute of Electrical and Electronics Engineers Inc., 2023. p. 938-943 (Proceedings - IEEE International Conference on Data Mining, ICDM).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Stochastic Integrated Explanations for Vision Models

AU - Barkan, Oren

AU - Elisha, Yehonatan

AU - Weill, Jonathan

AU - Asher, Yuval

AU - Eshel, Amit

AU - Koenigstein, Noam

PY - 2023

Y1 - 2023

N2 - We introduce Stochastic Integrated Explanations (SIX) - a general method for explaining predictions made by vision models. SIX employs stochastic integration on the internal representations across different network layers, producing explanation maps at various scales. The primary innovation of SIX is the introduction of randomness to the integration process by modeling the baseline representation as a random tensor. Through iterative sampling from the baseline distribution, SIX generates a diverse set of explanation maps, allowing the selection of the best-performing map based on a specific metric of interest. Extensive evaluations on various model architectures showcase the superior performance of SIX compared to state-of-the-art explanation methods, affirming its effectiveness across multiple metrics. Our code is available at: https://github.com/six-icdm/six

AB - We introduce Stochastic Integrated Explanations (SIX) - a general method for explaining predictions made by vision models. SIX employs stochastic integration on the internal representations across different network layers, producing explanation maps at various scales. The primary innovation of SIX is the introduction of randomness to the integration process by modeling the baseline representation as a random tensor. Through iterative sampling from the baseline distribution, SIX generates a diverse set of explanation maps, allowing the selection of the best-performing map based on a specific metric of interest. Extensive evaluations on various model architectures showcase the superior performance of SIX compared to state-of-the-art explanation methods, affirming its effectiveness across multiple metrics. Our code is available at: https://github.com/six-icdm/six

KW - Computer Vision

KW - Deep Learning

KW - Explainable AI

UR - http://www.scopus.com/inward/record.url?scp=85175602798&partnerID=8YFLogxK

U2 - 10.1109/ICDM58522.2023.00104

DO - 10.1109/ICDM58522.2023.00104

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:85175602798

T3 - Proceedings - IEEE International Conference on Data Mining, ICDM

SP - 938

EP - 943

BT - Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023

A2 - Chen, Guihai

A2 - Khan, Latifur

A2 - Gao, Xiaofeng

A2 - Qiu, Meikang

A2 - Pedrycz, Witold

A2 - Wu, Xindong

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 23rd IEEE International Conference on Data Mining, ICDM 2023

Y2 - 1 December 2023 through 4 December 2023

ER -

Barkan O, Elisha Y, Weill J, Asher Y, Eshel A, Koenigstein N. Stochastic Integrated Explanations for Vision Models. In Chen G, Khan L, Gao X, Qiu M, Pedrycz W, Wu X, editors, Proceedings - 23rd IEEE International Conference on Data Mining, ICDM 2023. Institute of Electrical and Electronics Engineers Inc. 2023. p. 938-943. (Proceedings - IEEE International Conference on Data Mining, ICDM). doi: 10.1109/ICDM58522.2023.00104

Stochastic Integrated Explanations for Vision Models

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this