Model-Agnostic Visual Explanations via Approximate Bilinear Models

Boris Joukovsky; Fawaz Sammani; Nikos Deligiannis

doi:10.1109/ICIP49359.2023.10222440

Model-Agnostic Visual Explanations via Approximate Bilinear Models

Boris Joukovsky, Fawaz Sammani, Nikos Deligiannis

Research output: Chapter in Book/Report/Conference proceeding › Conference paper

169 Downloads (Pure)

Abstract

This paper proposes InteractionLIME: a model-agnostic attribution technique to explain deep models predictions in terms of feature interactions. Specifically, we regress a bilinear form to approximate the output of two-input models, by sampling perturbations of both inputs simultaneously. Upon training, we retrieve a global explanation and a set of feature partitioning maps via the singular value decomposition of the learned interaction matrix of the bilinear model. We demonstrate InteractionLIME on vision and text-vision contrastive models, using visual examples and quantitative evaluation metrics. Our results show that the bilinear model successfully retrieves important interacting features from both inputs, while strongly reducing the occurrence of incomplete or asymmetric explanations produced by a linear model.

Original language	English
Title of host publication	2023 IEEE International Conference on Image Processing
Publisher	IEEE
Pages	1770-1774
Number of pages	5
ISBN (Electronic)	978-1-7281-9835-4
ISBN (Print)	978-1-7281-9836-1
DOIs	https://doi.org/10.1109/ICIP49359.2023.10222440
Publication status	Published - 11 Jul 2023
Event	2023 IEEE International Conference on Image Processing - Kuala Lumpur Convention Center (KLCC), Kuala Lumpur, Malaysia Duration: 8 Oct 2023 → 11 Oct 2023 https://2023.ieeeicip.org/

Publication series

Name	Proceedings - International Conference on Image Processing, ICIP
ISSN (Print)	1522-4880

Conference

Conference	2023 IEEE International Conference on Image Processing
Abbreviated title	ICIP2023
Country/Territory	Malaysia
City	Kuala Lumpur
Period	8/10/23 → 11/10/23
Internet address	https://2023.ieeeicip.org/

Bibliographical note

Funding Information:
This research received funding from the Flemish Government under the “Onderzoeksprogramma Artificiële Intelligentie (AI) Vlaanderen” programme, and from the FWO (Grants 1SB5721N and G0A4720N), Belgium.

Publisher Copyright:
© 2023 IEEE.

Access to Document

10.1109/ICIP49359.2023.10222440

InteractionLIME2023
© 2023 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 3.95 MBLicence: Unspecified

https://ieeexplore.ieee.org/abstract/document/10222440

Cite this

@inproceedings{47ef400d48d142fc826ee0859509c66d,

title = "Model-Agnostic Visual Explanations via Approximate Bilinear Models",

abstract = "This paper proposes InteractionLIME: a model-agnostic attribution technique to explain deep models predictions in terms of feature interactions. Specifically, we regress a bilinear form to approximate the output of two-input models, by sampling perturbations of both inputs simultaneously. Upon training, we retrieve a global explanation and a set of feature partitioning maps via the singular value decomposition of the learned interaction matrix of the bilinear model. We demonstrate InteractionLIME on vision and text-vision contrastive models, using visual examples and quantitative evaluation metrics. Our results show that the bilinear model successfully retrieves important interacting features from both inputs, while strongly reducing the occurrence of incomplete or asymmetric explanations produced by a linear model.",

author = "Boris Joukovsky and Fawaz Sammani and Nikos Deligiannis",

note = "Funding Information: This research received funding from the Flemish Government under the “Onderzoeksprogramma Artifici{\"e}le Intelligentie (AI) Vlaanderen” programme, and from the FWO (Grants 1SB5721N and G0A4720N), Belgium. Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE International Conference on Image Processing, ICIP2023 ; Conference date: 08-10-2023 Through 11-10-2023",

year = "2023",

month = jul,

day = "11",

doi = "10.1109/ICIP49359.2023.10222440",

language = "English",

isbn = "978-1-7281-9836-1",

series = "Proceedings - International Conference on Image Processing, ICIP",

publisher = "IEEE",

pages = "1770--1774",

booktitle = "2023 IEEE International Conference on Image Processing",

url = "https://2023.ieeeicip.org/",

}

Joukovsky, B , Sammani, F & Deligiannis, N 2023, Model-Agnostic Visual Explanations via Approximate Bilinear Models. in 2023 IEEE International Conference on Image Processing. Proceedings - International Conference on Image Processing, ICIP, IEEE, pp. 1770-1774, 2023 IEEE International Conference on Image Processing, Kuala Lumpur, Malaysia, 8/10/23. https://doi.org/10.1109/ICIP49359.2023.10222440

TY - GEN

T1 - Model-Agnostic Visual Explanations via Approximate Bilinear Models

AU - Joukovsky, Boris

AU - Sammani, Fawaz

AU - Deligiannis, Nikos

N1 - Funding Information: This research received funding from the Flemish Government under the “Onderzoeksprogramma Artificiële Intelligentie (AI) Vlaanderen” programme, and from the FWO (Grants 1SB5721N and G0A4720N), Belgium. Publisher Copyright: © 2023 IEEE.

PY - 2023/7/11

Y1 - 2023/7/11

N2 - This paper proposes InteractionLIME: a model-agnostic attribution technique to explain deep models predictions in terms of feature interactions. Specifically, we regress a bilinear form to approximate the output of two-input models, by sampling perturbations of both inputs simultaneously. Upon training, we retrieve a global explanation and a set of feature partitioning maps via the singular value decomposition of the learned interaction matrix of the bilinear model. We demonstrate InteractionLIME on vision and text-vision contrastive models, using visual examples and quantitative evaluation metrics. Our results show that the bilinear model successfully retrieves important interacting features from both inputs, while strongly reducing the occurrence of incomplete or asymmetric explanations produced by a linear model.

AB - This paper proposes InteractionLIME: a model-agnostic attribution technique to explain deep models predictions in terms of feature interactions. Specifically, we regress a bilinear form to approximate the output of two-input models, by sampling perturbations of both inputs simultaneously. Upon training, we retrieve a global explanation and a set of feature partitioning maps via the singular value decomposition of the learned interaction matrix of the bilinear model. We demonstrate InteractionLIME on vision and text-vision contrastive models, using visual examples and quantitative evaluation metrics. Our results show that the bilinear model successfully retrieves important interacting features from both inputs, while strongly reducing the occurrence of incomplete or asymmetric explanations produced by a linear model.

UR - http://www.scopus.com/inward/record.url?scp=85180791013&partnerID=8YFLogxK

U2 - 10.1109/ICIP49359.2023.10222440

DO - 10.1109/ICIP49359.2023.10222440

M3 - Conference paper

SN - 978-1-7281-9836-1

T3 - Proceedings - International Conference on Image Processing, ICIP

SP - 1770

EP - 1774

BT - 2023 IEEE International Conference on Image Processing

PB - IEEE

T2 - 2023 IEEE International Conference on Image Processing

Y2 - 8 October 2023 through 11 October 2023

ER -

Model-Agnostic Visual Explanations via Approximate Bilinear Models

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Embargoed Document

Embargoed Document

Fingerprint

VLAAI1: Subsidie: Onderzoeksprogramma Artificiële Intelligentie (AI) Vlaanderen

FWOSB97: Interpretable and Explainable Deep Learning for Video Processing

FWOAL972: Design and Interpret: A New Framework for Explainable AI