Special Session - [ExFMA-2026] Explainability and Fairness in Multimedia Analysis Call for Papers

==================================================================

22nd International Conference on Content-based Multimedia Indexing, CBMI 2026
 Toulouse, France, October 21-23, 2026

https://cbmi2026.sciencesconf.org/

==================================================================
[CBMI2026] Special Session - 
[ExFMA-2026] Explainability and Fairness in Multimedia Analysis

Recent advances in machine learning, and in particular deep learning,
have led to remarkable performance gains in multimedia analysis
tasks. However, it has also raised questions about the reliability,
explicability, and fairness of their predictions for decision-making
(e.g., the black box problem of the deep models and the risk of biased
outcomes). This lack of transparency and potential unfairness raises
many ethical and political concerns that prevent wider adoption of
this potentially highly beneficial technology, especially when such
systems are deployed in high-stakes or socially sensitive domains.

Most multimedia applications, such as person detection/tracking, face
recognition, or lifelog analysis, involve sensitive personal
information. This raises both legal issues, such as data protection
and regulations in the ongoing European AI regulation, as well as
ethical concerns related to discrimination, demographic bias, and
potential misuse of these technologies.

These challenges are particularly acute in multimedia applications,
where models operate on high-dimensional, multimodal data, and where
predictions frequently rely on subtle semantic cues that are difficult
to interpret even for human experts. Biases may emerge from data
imbalance, annotation practices, model design, or deployment contexts,
and may disproportionately affect certain individuals or
communities. It is therefore crucial not only to understand how
predictions correlate with information perception and expert
decision-making but also whether they are equitable across groups and
aligned with societal values. The objective of eXplainable AI (XAI)
and Fair AI is to improve transparency, mitigate bias, and foster
meaningful human

understanding of AI systems.

This special session focuses on methods and applications for
explainable and fair multimedia analysis, with an emphasis on
explanations that are faithful to the underlying models, meaningful to
end users, actionable for domain experts, and supportive of bias
detection and mitigation. The goal is to bring together researchers
and practitioners working on theoretical, methodological, and applied
aspects of explainability, fairness, evaluation, and interaction in
multimedia AI systems.

Topics of interest include (but are not limited to):

    Analysis of the influencing factors relevant to the final decision as an essential step to understand and improve the underlying processes involved.
    Methods for bias detection, fairness assessment, and mitigation in multimedia dataset and models.
    Fairness-aware learning strategies for multimedia analysis.
    Information visualization for models or their predictions.
    Visual analytics and Interactive applications for XAI.
    Performance evaluation metrics and protocols for explainability.
     Performance evaluation metrics and protocols for fairness.
    Sample-centric and dataset-centric explanations, including subgroup analyses
    Attention mechanisms for XAI.
    XAI-based pruning.
    XAI for multimedia systems supporting domain experts (e.g., healthcare, security, cultural heritage).
    Open challenges from industry or existing and emerging regulatory frameworks.
    Industrial use cases and deployment challenges.

The special session aims to collect high-quality scientific
contributions that advance the state of the art in explainable and
fair multimedia analysis, and to foster interdisciplinary discussion
on how transparency, fairness, and accountability can be jointly
addressed in multimedia AI systems. By integrating explainability and
fairness, the session seeks to promote trustworthy AI technologies
that enhance societal benefit while minimizing risks of bias,
discrimination, and unintended harm.

Important dates :

Paper deadline: 20 APRIL  2026 

Notification: 22 MAY  2026

Camera-ready: 15 JUNE 2026

Paper submission  : Author Guidelines

Please indicate in the comments that this paper is for SS  ExFMA-2026

SS chairs

    Chiara Galdi, EURECOM, Sophia Antipolis, France.
    Romain Bourqui, Université of Bordeaux
    Martin Winter, JOANNEUM RESEARCH - DIGITAL, Graz, Austria.
    Romain Giot, Université of Bordeaux