Multi-Attribute, Multimodal Bias Mitigation in AI Systems

Informazioni relative al progetto

MAMMOth

ID dell’accordo di sovvenzione: 101070285

DOI

10.3030/101070285

Progetto chiuso

Data della firma CE 20 Giugno 2022

Data di avvio 1 Novembre 2022

Data di completamento 31 Ottobre 2025

Finanziato da

• Digital, Industry and Space

Costo totale

€ 3 304 975,00

Contributo UE

€ 3 304 975,00

Coordinato da

ETHNIKO KENTRO EREVNAS KAI TECHNOLOGIKIS ANAPTYXIS
Greece

CORDIS fornisce collegamenti ai risultati finali pubblici e alle pubblicazioni dei progetti ORIZZONTE.

I link ai risultati e alle pubblicazioni dei progetti del 7° PQ, così come i link ad alcuni tipi di risultati specifici come dataset e software, sono recuperati dinamicamente da .OpenAIRE .

Risultati finali

Documents, reports (11)

Explainable Methods for Discovering Bias in Multimodal Data

Report on outcomes of activities T2.3 and T2.4, focusing on the interaction between multimodal data and multi-attribute network analysis streams and the development of explainable bias analysis methods.

Methods and Algorithms for Multidimensional Fairness in Multi-attribute Networks

Report on research for mitigating multi-dimensional discrimination in multi-attribute networks.

Methods and Algorithms for Multidimensional Fairness-aware Classification

Report on methods for mitigating bias in multi-dimensional classification problems.

Multi-dimensional Discrimination Definitions and Operational Measures

Report on operational definitions and measures for multi-dimensional fairness.

Stakeholder Evaluation of System, Processes and Results

This summarizes and prioritizes the results of the three pilots, along with the assessment of the reliability of their methodological aspects. The report will further suggest measures that could form the basis for changes in formal legislation.

MAMMOth's Handbook: Best practices

Consolidated training and awareness raising material on preventing gender and intersectional bias for AI researchers, students and practitioners.

Final DCE and Policy Report

Report containing updates on DCE activities (CSI), standardization (CSI) and policy advice (TRI).

Methods and Algorithms for Bias-aware Multimodal attribute Extraction

Report on methods for assessing the bias of AI multimodal attribute extraction models and on improved models that do not discriminate against protected or sensitive groups.

Methods and Algorithms for Bias Analysis in Multi-attribute Networks

Report on methods for bias analysis and modeling multi-attribute networks.

User Requirements and Architecture

This report will document the activities, methodology and outcomes of the co-creation and analysis of system/user requirements and will outline the overall architecture of the MAMMoth bias toolkit. UNIBO will be in charge of the user requirements analysis, EXUS will lead the toolkit architecture design, and CERTH will coordinate the overall process and writing of the deliverable.

Evaluation and Mitigation of Bias in the use cases

This summarizes the findings of applying the MAMMOth toolkit and methods on the three pilots.

Other (1)

MAMMOth Bias Toolkit

Software toolkit for bias exploration, evaluation and mitigation. A first and second version of the toolkit will be released on M18 (MS4) and M26 (MS5) respectively.

Data Management Plan (1)

Data Management Plan

First version of the Data Management Plan. It will be maintained and updated internally throughout the project.

Pubblicazioni

Conference proceedings (19)

hyperFA*IR: A hypergeometric approach to fair rankings with finite candidate pool

Autori: Mauritz N. Cartier van Dissel, Samuel Martin-Gutierrez, Lisette Espín-Noboa, Ana María Jaramillo, Fariba Karimi
Pubblicato in: Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, 2025
Editore: ACM
DOI: 10.1145/3715275.3732143

FRCSyn Challenge at WACV 2024: Face Recognition Challenge in the Era of Synthetic Data

Autori: Pietro Melzi, Ruben Tolosana, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Ivan DeAndres-Tame, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, Weisong Zhao, Xiangyu Zhu, Zheyu Yan, Xiao-Yu Zhang, Jinlin Wu, Zhen Lei, Suvidha
Pubblicato in: FRCSyn Challenge at WACV 2024: Face Recognition Challenge in the Era of Synthetic Data
Editore: Elsevier BV
DOI: 10.1109/WACVW60836.2024.00100

MMM-fair: An Interactive Toolkit for Exploring and Operationalizing Multi-Fairness Trade-offs

Autori: Swati Swati, Arjun Roy, Emmanouil Panagiotou, Eirini Ntoutsi
Pubblicato in: Proceedings of the 34th ACM International Conference on Information and Knowledge Management, 2025
Editore: ACM
DOI: 10.1145/3746252.3761476

Badd: Bias mitigation through bias addition

Autori: Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, Christos Diou
Pubblicato in: 2025
Editore: IEEE/CVF

Studying bias in visual features through the lens of optimal transport.

Autori: Simone Fabbrizzi, Xuan Zhao, Emmanouil Krasanakis, Symeon Papadopoulos, and Eirini Ntoutsi
Pubblicato in: Studying bias in visual features through the lens of optimal transport., 2023
Editore: Data Mining and Knowledge Discovery
DOI: 10.1007/s10618-023-00972-2

FairBranch: Mitigating Bias Transfer in Fair Multi-task Learning

Autori: Arjun Roy, Christos Koutlis, Symeon Papadopoulos, Eirini Ntoutsi
Pubblicato in: 2024 International Joint Conference on Neural Networks (IJCNN), 2024
Editore: IEEE
DOI: 10.1109/IJCNN60899.2024.10651221

Face-swapping based Data Augmentation for ID Document and Selfie Face Verification

Autori: Elmokhtar Mohamed Moussa, Ioannis Sarridis, Emmanouil Krasanakis, Nathan Ramoly, Symeon Papadopoulos, Ahmad Montaser Awal, Lara Younes
Pubblicato in: Proceedings of the Winter Conference on Applications of Computer Vision (WACV) Workshops, Numero pp. 1421-1428, 2025
Editore: CVF

A Deep Latent Factor Graph Clustering with Fairness-Utility Trade-off Perspective

Autori: Siamak Ghodsi, Amjad Seyedi, Tai Le Quy, Fariba Karimi, and Eirini Ntoutsi
Editore: IEEE

Towards Cohesion-Fairness Harmony: Contrastive Regularization in Individual Fair Graph Clustering

Autori: Ghodsi, S., Amjad Seyedi, S., and Ntoutsi, E
Pubblicato in: Towards Cohesion-Fairness Harmony: Contrastive Regularization in Individual Fair Graph Clustering, 2024, ISBN 978-981-97-2242-6
Editore: Springer, Singapore
DOI: 10.1007/978-981-97-2242-6_23

Exploring Fusion Techniques in Multimodal AI-Based Recruitment: Insights from FairCVdb

Autori: """Swati, Swati, Arjun Roy, Eirini Ntoutsi """
Pubblicato in: 2024
Editore: EWAF 2024
DOI: 10.48550/ARXIV.2407.16892

Multi-dimensional Discrimination in Law and Machine Learning - A Comparative Overview

Autori: Arjun Roy, Jan Horstmann, Eirini Ntoutsi
Pubblicato in: 2023 ACM Conference on Fairness, Accountability, and Transparency, 2025
Editore: ACM
DOI: 10.1145/3593013.3593979

Adversarial Robustness of VAEs across Intersectional Subgroups

Autori: Ramanaik, Chethan Krishnamurthy and Roy, Arjun and Ntoutsi, Eirini
Pubblicato in: 2024
Editore: Accepted in the BIAS workshop, co-located with ECML PKDD 2024
DOI: 10.48550/ARXIV.2407.03864

FaceX: Understanding Face Attribute Classifiers through Summary Model Explanations

Autori: Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, Christos Diou
Pubblicato in: Proceedings of the 2024 International Conference on Multimedia Retrieval, 2025
Editore: ACM
DOI: 10.1145/3652583.3658007

Synthetic Tabular Data Generation for Class Imbalance and Fairness: A Comparative Study

Autori: Panagiotou, Emmanouil and Roy, Arjun and Ntoutsi, Eirini
Pubblicato in: 2024
Editore: Accepted in the BIAS workshop, co-located with ECML PKDD 2024
DOI: 10.48550/ARXIV.2409.05215

Achieving Socio-Economic Parity through the Lens of EU AI Act

Autori: Arjun Roy, Stavroula Rizou, Symeon Papadopoulos, Eirini Ntoutsi
Pubblicato in: Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency, 2025, ISBN 9798400714825
Editore: ACM
DOI: 10.1145/3715275.3732125

Learning impartial policies for sequential counterfactual explanations using Deep Reinforcement Learning

Autori: Emmanouil Panagiotou, Eirini Ntoutsi
Pubblicato in: 2025
Editore: Springer, Cham
DOI: 10.1007/978-3-031-74630-7_24

SDFD: Building a Versatile Synthetic Face Image Dataset with Diverse Attributes

Autori: Georgia Baltsou, Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos
Pubblicato in: 2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG), 2024
Editore: IEEE
DOI: 10.1109/FG59268.2024.10581864

Towards Standardizing AI Bias Exploration

Autori: Emmanouil Krasanakis, Symeon Papadopoulos
Pubblicato in: Workshop on AI bias: Measurements, Mitigation, Explanation Strategies (AIMMES 2024), 2024, ISSN 2331-8422
Editore: Workshop on AI bias: Measurements, Mitigation, Explanation Strategies (AIMMES 2024)
DOI: 10.48550/arXiv.2405.19022

TABCF: Counterfactual Explanations for Tabular Data Using a Transformer-Based VAE

Autori: Emmanouil Panagiotou, Manuel Heurich, Tim Landgraf, Eirini Ntoutsi
Pubblicato in: Proceedings of the 5th ACM International Conference on AI in Finance, 2024
Editore: ACM
DOI: 10.1145/3677052.3698673

Peer reviewed articles (12)

Training in Co-Creation as a Methodological Approach to Improve AI Fairness

Autori: Ian Slesinger, Evren Yalaz, Stavroula Rizou, Marta Gibin, Emmanouil Krasanakis, Symeon Papadopoulos
Pubblicato in: Societies, Numero 14, 2024, ISSN 2075-4698
Editore: MDPI AG
DOI: 10.3390/SOC14120259

Universal Local Attractors on Graphs

Autori: Emmanouil Krasanakis, Symeon Papadopoulos, Ioannis Kompatsiaris
Pubblicato in: Applied Sciences, Numero 14, 2024, ISSN 2076-3417
Editore: MDPI AG
DOI: 10.3390/APP14114533

FRCSyn-onGoing: Benchmarking and comprehensive evaluation of real and synthetic data to improve face recognition systems

Autori: Pietro Melzi, Ruben Tolosana, Ruben Vera-Rodriguez, Minchul Kim, Christian Rathgeb, Xiaoming Liu, Ivan DeAndres-Tame, Aythami Morales, Julian Fierrez, Javier Ortega-Garcia, Weisong Zhao, Xiangyu Zhu, Zheyu Yan, Xiao-Yu Zhang, Jinlin Wu, Zhen Lei, Suvidha
Pubblicato in: FRCSyn-onGoing: Benchmarking and comprehensive evaluation of real and synthetic data to improve face recognition systems, 2024, ISSN 1566-2535
Editore: Elsevier
DOI: 10.1016/j.inffus.2024.102322

Negotiating AI fairness: a call for rebalancing power relations

Autori: Marta Gibin, Emmanouil Krasanakis, Kęstutis Grumodas
Pubblicato in: AI & SOCIETY, 2025, ISSN 0951-5666
Editore: Springer Science and Business Media LLC
DOI: 10.1007/S00146-025-02653-8

Towards Fair Face Verification: An In-depth Analysis of Demographic Biases

Autori: Ioannis Sarridis; Christos Koutlis; Symeon Papadopoulos; Christos Diou
Pubblicato in: Proceedings of BIAS 2023 Workshop, ECML/PKDD, 2023, ISSN 2331-8422
Editore: Proceedings of BIAS 2023 Workshop, ECML/PKDD
DOI: 10.48550/arxiv.2307.10011

FLAC: Fairness-Aware Representation Learning by Suppressing Attribute-Class Associations

Autori: Ioannis Sarridis, Christos Koutlis, Symeon Papadopoulos, Christos Diou
Pubblicato in: IEEE Transactions on Pattern Analysis and Machine Intelligence, Numero 47, 2024, ISSN 0162-8828
Editore: Institute of Electrical and Electronics Engineers (IEEE)
DOI: 10.1109/TPAMI.2024.3487254

Can we Disrupt the Momentum of the AI Colonization of Science Education?

Autori: Lucy Avraamidou
Pubblicato in: Can we Disrupt the Momentum of the AI Colonization of Science Education?, 2024, ISSN 1098-2736
Editore: Journal of Research in Science Teaching
DOI: 10.1002/tea.21961

Assortative and preferential attachment lead to core-periphery networks

Autori: Javier Ureña-Carrión, Fariba Karimi, Gerardo Íñiguez, Mikko Kivelä
Pubblicato in: Physical Review Research, Numero 5, 2023, ISSN 2643-1564
Editore: American Physical Society (APS)
DOI: 10.1103/PHYSREVRESEARCH.5.043287

The Use of AI in school science: a Systematic Literature Review

Autori: Dagmar Mercedes Heeg, Lucy Avraamidou
Pubblicato in: The Use of AI in school science: a Systematic Literature Review, 2023, ISSN 1469-5790
Editore: Taylor & Francis Group
DOI: 10.1080/09523987.2023.2264990

On the inadequacy of nominal assortativity for assessing homophily in networks

Autori: Fariba Karimi, Marcos Oliveira
Pubblicato in: Scientific Reports, Numero 13, 2024, ISSN 2045-2322
Editore: Springer Science and Business Media LLC
DOI: 10.1038/S41598-023-48113-5

Young children’s understanding of AI

Autori: Dagmar Mercedes Heeg, Lucy Avraamidou
Pubblicato in: Education and Information Technologies, 2024, ISSN 1573-7608
Editore: Springer Nature
DOI: 10.1007/S10639-024-13169-X

Improving the visibility of minorities through network growth interventions

Autori: Leonie Neuhäuser, Fariba Karimi, Jan Bachmann, Markus Strohmaier, Michael T Schaub
Pubblicato in: Improving the visibility of minorities through network growth interventions, 2023, ISSN 2399-3650
Editore: Communications Physics
DOI: 10.1038/s42005-023-01218-9

Book chapters (1)

Sampling Strategies for Mitigating Bias in Face Synthesis Methods

Autori: Emmanouil Maragkoudakis, Symeon Papadopoulos, Iraklis Varlamis, Christos Diou
Pubblicato in: Communications in Computer and Information Science, Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2025
Editore: Springer Nature Switzerland
DOI: 10.1007/978-3-031-74630-7_16