Visual Explanations via Iterated Integrated Attributions

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

doi:10.1109/ICCV51070.2023.00198

Visual Explanations via Iterated Integrated Attributions

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

פרסום מחקרי: פרק בספר / בדוח / בכנס › פרסום בספר כנס › ביקורת עמיתים

תקציר

We introduce Iterated Integrated Attributions (IIA) - a generic method for explaining the predictions of vision models. IIA employs iterative integration across the input image, the internal representations generated by the model, and their gradients, yielding precise and focused explanation maps. We demonstrate the effectiveness of IIA through comprehensive evaluations across various tasks, datasets, and network architectures. Our results showcase that IIA produces accurate explanation maps, outperforming other state-of-the-art explanation techniques.

שפה מקורית	אנגלית
כותר פרסום המארח	Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023
מוציא לאור	Institute of Electrical and Electronics Engineers Inc.
עמודים	2073-2084
מספר עמודים	12
מסת"ב (אלקטרוני)	9798350307184
מזהי עצם דיגיטלי (DOIs)	https://doi.org/10.1109/ICCV51070.2023.00198
סטטוס פרסום	פורסם - 2023
פורסם באופן חיצוני	כן
אירוע	2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023 - Paris, צרפת משך הזמן: 2 אוק׳ 2023 → 6 אוק׳ 2023

סדרות פרסומים

שם	Proceedings of the IEEE International Conference on Computer Vision
ISSN (מודפס)	1550-5499

כנס

כנס	2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023
מדינה/אזור	צרפת
עיר	Paris
תקופה	2/10/23 → 6/10/23

הערה ביבליוגרפית

Publisher Copyright:
© 2023 IEEE.

גישה למסמך

10.1109/ICCV51070.2023.00198

קבצים וקישורים אחרים

Link to publication in Scopus

פורמט ציטוט ביבליוגרפי

Barkan, O., Elisha, Y., Asher, Y., Eshel, A., & Koenigstein, N. (2023). Visual Explanations via Iterated Integrated Attributions. ב- Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023 (עמודים 2073-2084). (Proceedings of the IEEE International Conference on Computer Vision). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCV51070.2023.00198

@inproceedings{d3bac99229614ac1ae7aceb5e0a13cad,

title = "Visual Explanations via Iterated Integrated Attributions",

abstract = "We introduce Iterated Integrated Attributions (IIA) - a generic method for explaining the predictions of vision models. IIA employs iterative integration across the input image, the internal representations generated by the model, and their gradients, yielding precise and focused explanation maps. We demonstrate the effectiveness of IIA through comprehensive evaluations across various tasks, datasets, and network architectures. Our results showcase that IIA produces accurate explanation maps, outperforming other state-of-the-art explanation techniques.",

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

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023 ; Conference date: 02-10-2023 Through 06-10-2023",

year = "2023",

doi = "10.1109/ICCV51070.2023.00198",

language = "אנגלית",

series = "Proceedings of the IEEE International Conference on Computer Vision",

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

pages = "2073--2084",

booktitle = "Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023",

address = "ארצות הברית",

}

Barkan, O, Elisha, Y, Asher, Y, Eshel, A & Koenigstein, N 2023, Visual Explanations via Iterated Integrated Attributions. ב- Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023. Proceedings of the IEEE International Conference on Computer Vision, Institute of Electrical and Electronics Engineers Inc., עמודים 2073-2084, 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023, Paris, צרפת, 2/10/23. https://doi.org/10.1109/ICCV51070.2023.00198

Visual Explanations via Iterated Integrated Attributions. / Barkan, Oren; Elisha, Yehonatan; Asher, Yuval et al.
Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023. Institute of Electrical and Electronics Engineers Inc., 2023. עמוד 2073-2084 (Proceedings of the IEEE International Conference on Computer Vision).

פרסום מחקרי: פרק בספר / בדוח / בכנס › פרסום בספר כנס › ביקורת עמיתים

TY - GEN

T1 - Visual Explanations via Iterated Integrated Attributions

AU - Barkan, Oren

AU - Elisha, Yehonatan

AU - Asher, Yuval

AU - Eshel, Amit

AU - Koenigstein, Noam

PY - 2023

Y1 - 2023

N2 - We introduce Iterated Integrated Attributions (IIA) - a generic method for explaining the predictions of vision models. IIA employs iterative integration across the input image, the internal representations generated by the model, and their gradients, yielding precise and focused explanation maps. We demonstrate the effectiveness of IIA through comprehensive evaluations across various tasks, datasets, and network architectures. Our results showcase that IIA produces accurate explanation maps, outperforming other state-of-the-art explanation techniques.

AB - We introduce Iterated Integrated Attributions (IIA) - a generic method for explaining the predictions of vision models. IIA employs iterative integration across the input image, the internal representations generated by the model, and their gradients, yielding precise and focused explanation maps. We demonstrate the effectiveness of IIA through comprehensive evaluations across various tasks, datasets, and network architectures. Our results showcase that IIA produces accurate explanation maps, outperforming other state-of-the-art explanation techniques.

UR - http://www.scopus.com/inward/record.url?scp=85175598274&partnerID=8YFLogxK

U2 - 10.1109/ICCV51070.2023.00198

DO - 10.1109/ICCV51070.2023.00198

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

AN - SCOPUS:85175598274

T3 - Proceedings of the IEEE International Conference on Computer Vision

SP - 2073

EP - 2084

BT - Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023

Y2 - 2 October 2023 through 6 October 2023

ER -

Barkan O, Elisha Y, Asher Y, Eshel A, Koenigstein N. Visual Explanations via Iterated Integrated Attributions. ב- Proceedings - 2023 IEEE/CVF International Conference on Computer Vision, ICCV 2023. Institute of Electrical and Electronics Engineers Inc. 2023. עמוד 2073-2084. (Proceedings of the IEEE International Conference on Computer Vision). doi: 10.1109/ICCV51070.2023.00198

Visual Explanations via Iterated Integrated Attributions

תקציר

סדרות פרסומים

כנס

הערה ביבליוגרפית

גישה למסמך

קבצים וקישורים אחרים

טביעת אצבע

פורמט ציטוט ביבליוגרפי