Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI

Qi Huang; Emanuele Mezzi; Osman Mutlu; Miltiadis Kofinas; Vidya Prasad; Shadnan Azwad Khan; Elena Ranguelova; Niki van Stein

doi:10.1007/978-3-031-63787-2_16

Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI

Qi Huang, Emanuele Mezzi, Osman Mutlu, Miltiadis Kofinas, Vidya Prasad, Shadnan Azwad Khan, Elena Ranguelova, Niki van Stein^*

^*Corresponding author for this work

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

Abstract

We introduce a novel metric for measuring semantic continuity in Explainable AI methods and machine learning models. We posit that for models to be truly interpretable and trustworthy, similar inputs should yield similar explanations, reflecting a consistent semantic understanding. By leveraging XAI techniques, we assess semantic continuity in the task of image recognition. We conduct experiments to observe how incremental changes in input affect the explanations provided by different XAI methods. Through this approach, we aim to evaluate the models’ capability to generalize and abstract semantic concepts accurately and to evaluate different XAI methods in correctly capturing the model behaviour. This paper contributes to the broader discourse on AI interpretability by proposing a quantitative measure for semantic continuity for XAI methods, offering insights into the models’ and explainers’ internal reasoning processes, and promoting more reliable and transparent AI systems.

Original language	English
Title of host publication	Explainable Artificial Intelligence
Subtitle of host publication	2nd World Conference, xAI 2024, Proceedings
Editors	Luca Longo, Sebastian Lapuschkin, Christin Seifert
Publisher	Springer
Pages	308-331
Number of pages	24
ISBN (Print)	9783031637865
DOIs	https://doi.org/10.1007/978-3-031-63787-2_16
Publication status	Published - 2024
Event	2nd World Conference on Explainable Artificial Intelligence, xAI 2024 - Valletta, Malta Duration: 17 Jul 2024 → 19 Jul 2024

Publication series

Name	Communications in Computer and Information Science
Volume	2153
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference/symposium

Conference/symposium	2nd World Conference on Explainable Artificial Intelligence, xAI 2024
Country/Territory	Malta
City	Valletta
Period	17/07/24 → 19/07/24

Keywords

Explainable AI
Machine Learning Interpretability
Semantic Analysis
Semantic Continuity

Access to Document

10.1007/978-3-031-63787-2_16

EU-23033 EFRA (KB-50-005-007)
Hürriyetoğlu, A. (Project Leader)
1/01/24 → 31/12/24
Project: LVVN project
EU23033 - EFRA (BO-64-101-014)
van der Velden, B. (Project Leader)
1/01/23 → 31/12/23
Project: LVVN project

Cite this

Huang, Q., Mezzi, E., Mutlu, O., Kofinas, M., Prasad, V., Khan, S. A., Ranguelova, E., & van Stein, N. (2024). Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI. In L. Longo, S. Lapuschkin, & C. Seifert (Eds.), Explainable Artificial Intelligence : 2nd World Conference, xAI 2024, Proceedings (pp. 308-331). (Communications in Computer and Information Science; Vol. 2153). Springer. https://doi.org/10.1007/978-3-031-63787-2_16

@inproceedings{cb36782c49344578b8d5675989fc99f3,

title = "Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI",

abstract = "We introduce a novel metric for measuring semantic continuity in Explainable AI methods and machine learning models. We posit that for models to be truly interpretable and trustworthy, similar inputs should yield similar explanations, reflecting a consistent semantic understanding. By leveraging XAI techniques, we assess semantic continuity in the task of image recognition. We conduct experiments to observe how incremental changes in input affect the explanations provided by different XAI methods. Through this approach, we aim to evaluate the models{\textquoteright} capability to generalize and abstract semantic concepts accurately and to evaluate different XAI methods in correctly capturing the model behaviour. This paper contributes to the broader discourse on AI interpretability by proposing a quantitative measure for semantic continuity for XAI methods, offering insights into the models{\textquoteright} and explainers{\textquoteright} internal reasoning processes, and promoting more reliable and transparent AI systems.",

keywords = "Explainable AI, Machine Learning Interpretability, Semantic Analysis, Semantic Continuity",

author = "Qi Huang and Emanuele Mezzi and Osman Mutlu and Miltiadis Kofinas and Vidya Prasad and Khan, {Shadnan Azwad} and Elena Ranguelova and {van Stein}, Niki",

year = "2024",

doi = "10.1007/978-3-031-63787-2_16",

language = "English",

isbn = "9783031637865",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "308--331",

editor = "Luca Longo and Sebastian Lapuschkin and Christin Seifert",

booktitle = "Explainable Artificial Intelligence",

address = "Germany",

note = "2nd World Conference on Explainable Artificial Intelligence, xAI 2024 ; Conference date: 17-07-2024 Through 19-07-2024",

}

Huang, Q, Mezzi, E, Mutlu, O, Kofinas, M, Prasad, V, Khan, SA, Ranguelova, E & van Stein, N 2024, Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI. in L Longo, S Lapuschkin & C Seifert (eds), Explainable Artificial Intelligence : 2nd World Conference, xAI 2024, Proceedings. Communications in Computer and Information Science, vol. 2153, Springer, pp. 308-331, 2nd World Conference on Explainable Artificial Intelligence, xAI 2024, Valletta, Malta, 17/07/24. https://doi.org/10.1007/978-3-031-63787-2_16

Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI. / Huang, Qi; Mezzi, Emanuele; Mutlu, Osman et al.
Explainable Artificial Intelligence : 2nd World Conference, xAI 2024, Proceedings. ed. / Luca Longo; Sebastian Lapuschkin; Christin Seifert. Springer, 2024. p. 308-331 (Communications in Computer and Information Science; Vol. 2153).

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

TY - GEN

T1 - Beyond the Veil of Similarity

T2 - 2nd World Conference on Explainable Artificial Intelligence, xAI 2024

AU - Huang, Qi

AU - Mezzi, Emanuele

AU - Mutlu, Osman

AU - Kofinas, Miltiadis

AU - Prasad, Vidya

AU - Khan, Shadnan Azwad

AU - Ranguelova, Elena

AU - van Stein, Niki

PY - 2024

Y1 - 2024

N2 - We introduce a novel metric for measuring semantic continuity in Explainable AI methods and machine learning models. We posit that for models to be truly interpretable and trustworthy, similar inputs should yield similar explanations, reflecting a consistent semantic understanding. By leveraging XAI techniques, we assess semantic continuity in the task of image recognition. We conduct experiments to observe how incremental changes in input affect the explanations provided by different XAI methods. Through this approach, we aim to evaluate the models’ capability to generalize and abstract semantic concepts accurately and to evaluate different XAI methods in correctly capturing the model behaviour. This paper contributes to the broader discourse on AI interpretability by proposing a quantitative measure for semantic continuity for XAI methods, offering insights into the models’ and explainers’ internal reasoning processes, and promoting more reliable and transparent AI systems.

AB - We introduce a novel metric for measuring semantic continuity in Explainable AI methods and machine learning models. We posit that for models to be truly interpretable and trustworthy, similar inputs should yield similar explanations, reflecting a consistent semantic understanding. By leveraging XAI techniques, we assess semantic continuity in the task of image recognition. We conduct experiments to observe how incremental changes in input affect the explanations provided by different XAI methods. Through this approach, we aim to evaluate the models’ capability to generalize and abstract semantic concepts accurately and to evaluate different XAI methods in correctly capturing the model behaviour. This paper contributes to the broader discourse on AI interpretability by proposing a quantitative measure for semantic continuity for XAI methods, offering insights into the models’ and explainers’ internal reasoning processes, and promoting more reliable and transparent AI systems.

KW - Explainable AI

KW - Machine Learning Interpretability

KW - Semantic Analysis

KW - Semantic Continuity

U2 - 10.1007/978-3-031-63787-2_16

DO - 10.1007/978-3-031-63787-2_16

M3 - Conference paper

AN - SCOPUS:85200793618

SN - 9783031637865

T3 - Communications in Computer and Information Science

SP - 308

EP - 331

BT - Explainable Artificial Intelligence

A2 - Longo, Luca

A2 - Lapuschkin, Sebastian

A2 - Seifert, Christin

PB - Springer

Y2 - 17 July 2024 through 19 July 2024

ER -

Huang Q, Mezzi E, Mutlu O, Kofinas M, Prasad V, Khan SA et al. Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI. In Longo L, Lapuschkin S, Seifert C, editors, Explainable Artificial Intelligence : 2nd World Conference, xAI 2024, Proceedings. Springer. 2024. p. 308-331. (Communications in Computer and Information Science). doi: 10.1007/978-3-031-63787-2_16

Beyond the Veil of Similarity: Quantifying Semantic Continuity in Explainable AI

Abstract

Publication series

Conference/symposium

Keywords

Access to Document

Fingerprint

Projects

EU-23033 EFRA (KB-50-005-007)

EU23033 - EFRA (BO-64-101-014)

Cite this