Salvatore Greco – M. Eugenia Cornejo – Xiaoling Wang – Didier Dubois – Francisco Herrera

Salvatore Greco is full professor at the Department of Economics and Business at the University of Catania where has been teaching Decision Theory, General Mathematics, Financial Mathematics and Actuarial Mathematics. He was the coordinator of the PhD in Mathematics for Decisions in Economics and Finance, and he is currently coordinator of the PhD in Economics, Management and Decision Making. His research regards preference modeling and multiple criteria decision analysis (MCDA) with a specific attention to application of rough set theory, robust ordinal regression, non-additive integrals, evolutionary multiobjective optimization methodologies, composite indices for sustainable development, wellbeing and innovation, MCDA models for territorial and urban planning.
At the 22nd International conference on MCDM (Multiple Criteria Decision Making) held in Malaga June 17-22 2013, he received the MCDM Gold Medal being “the highest honor that the International Society on Multiple Criteria Decision Making bestows upon a scholar who, over a distinguished career, has devoted much of his/her talent, time, and energy to advancing the field of MCDM, and who has markedly contributed to the theory, methodology, and practice of MCDM”.
Since 2010, Salvatore Greco is one of the three coordinators of the EURO Working Group in Multiple Criteria Decision Aiding. He has been member of the executive committee of International Society on Multiple Criteria Decision Making for the years 2006-2009, 2011-2013, 2016-2019, 2020-2024. In the years 2014-2019 Salvatore Greco was member of the scientific committee of AMASES (Italian Society for Mathematics Applied to Social and Economic Sciences) and in the years 2017-2019 he served as vicepresident. He was the president of the MCDM section of INFORMS for the 2022-2023 term. Salvatore Greco is the editor in chief of Decisions in Economics and Finance and area editor of Journal of Multicriteria Decision Analysis and Soft Computing. He is also in the editorial board of many scientific journals in the domain of operational research and decision analysis (e.g. European Journal of Operational Research, Fuzzy Sets and Systems, EURO Journal on Decision Processes).
Scopus reports 273 publications of Salvatore Greco cited all together 12286 times and an h-index of 59. Google Scholar reports a total of 29136 citations with an h-index of 79. Salvatore Greco is one of World’s Top 2% Scientists, included in the global list released by Stanford University in various disciplines since first edition in 2019.

Title: Knowing What We Don’t Know: Rough Set Approximation in Seven-valued Logic

Abstract: Knowing what we do not know is a central issue in reasoning under uncertainty. In this talk, we propose an extension of the classical rough set approximation that explicitly allows for the possibility that elements of the universe, besides being assigned or not assigned to a target class, may also be classified as unknown when no definite assignment can be made. This perspective makes it possible to distinguish between vagueness, arising from intrinsic indeterminacy in class membership, and ambiguity, resulting from the coarseness of indiscernibility classes that contain objects both belonging and not belonging to the target class. Within this framework, a general seven-valued logic naturally emerges, yielding a more refined representation of knowledge states. Interestingly, this logic corresponds to a reasoning system developed by Jaina philosophers as early as the 4th century BC. We show that rough set theory, originally introduced by Zdzisław Pawlak, can be consistently extended in this setting, and that the proposed model generalizes Nuel Belnap’s four-valued logic. Finally, we provide an algebraic foundation for this framework in the form of a Pawlak-Brouwer-Zadeh lattice.

M. Eugenia Cornejo is Assistant Professor at the Department of Mathematics, University of Cádiz, Spain. Her research areas of interest are fuzzy sets, fuzzy logic, logic programming, relational data analysis, fuzzy relation equations and algebraic structures for soft computing. She co-leads two national projects, a working group of the COST Action (DigForASP) and she has participated as a researcher in different national projects and research contracts with companies. She has published more than 150 papers in conference proceedings and scientific journals, including the Journal of Computational and Applied Mathematics, Fuzzy Sets and Systems, Information Sciences, International Journal of Intelligent Systems, International Journal of Approximate Reasoning Fuzzy Sets and Systems, and Transactions on Fuzzy Systems.

Title: Decision Making in Fuzzy Rough Set Theory

Abstract: This talk addresses the extension of decision algorithms to the fuzzy rough set theory framework, where datasets are modeled as decision tables under uncertainty. The notion of decision algorithm, originally defined by Pawlak as a collection of rules with desirable descriptive properties for relational systems, is revisited and adapted through the notion of efficiency. In this context, two approaches to defining efficiency in the fuzzy framework are presented: the first one is a direct generalization to the classical case, while the second one is focused on obtaining a bounded efficiency preserving the philosophy of the classical framework.

In addition, this talk presents different methods for classifying new objects which are not included in the original dataset, aiming to support more effective decision-making. These methods rely on relevance indicators associated with decision rules—such as support, certainty, and credibility—taking into account both the degree of matching between new objects and existing rules, as well as their representativeness and reliability. The proposed classification strategies extend and generalize existing approaches in the literature, offering richer and more flexible alternatives within the fuzzy rough set framework.

Xiaoling Wang is a professor and PhD supervisor at the School of Computer Science, East China Normal University. She is a senior member of the China Computer Federation (CCF) and an executive member of the Database Committee. Xiaoling Wang has long been engaged in research in big data, knowledge discovery, and AI. He has published over 150 papers in important international journals and prestigious academic conferences. She has applied for more than 20 national invention patents, with 18 already granted. She has led or attended key R&D projects from the Ministry of Science and Technology, projects from the National Natural Science Foundation of China. Her students have also achieved top placements in academic competitions at leading international conferences, including 12 excellent conference papers.

Title: Towards General Embodied AI: Spatio-Temporal Visual Reasoning from Egocentric Perspective

Abstract: Embodied AI demands that intelligent agents perceive, understand, and reason about the physical world from egocentric visual observations, which is a core step toward general embodied intelligence and differs fundamentally from conventional third-person visual understanding tasks. Egocentric visual reasoning poses unique and intractable challenges, including a constrained field of view, inherent visual uncertainty (e.g., occlusion, motion blur), severe cross-domain distribution shifts, and complex spatio-temporal dependencies in long-horizon context. These inherent properties render existing vision-language models (VLMs) inadequate to achieve stable scene understanding and reliable real-time reasoning in practical environments, due to their deficiencies in spatio-temporal representation, uncertainty handling, and streaming information processing.
In this talk, we present a comprehensive series of research efforts on egocentric spatio-temporal visual reasoning, addressing the above challenges from both data construction and methodological innovation perspectives. On the data side, we focus on building realistic, fine-grained, and task-oriented benchmarks along with scalable data construction pipelines for real-world embodied scenarios. Specifically, we introduce EgoCross (AAAI 2026), the first benchmark for cross-domain egocentric video question answering (VQA) that covers diverse domains including surgery, industry, extreme sports and animal perspectives, and StreamEQA (CVPR 2026 Findings), a pioneering embodied streaming VQA benchmark that decomposes reasoning tasks into perceptual, interactive and planning levels across backward, real-time and forward temporal dimensions.
On the methodological side, we propose structured spatio-temporal scene representation frameworks to bridge the critical capability gaps of VLMs in egocentric perception-reasoning and 2D semantic-3D geometric understanding. We first introduce CLiViS (CVPR 2026), which unleashes cognitive maps via linguistic-visual synergy to model spatio-temporal correlations between entities explicitly, enabling the fusion of low-level visual perception and high-level logical reasoning. We then present HSGM (CVPR 2026), a hierarchical semantic-geometric map that decomposes 3D physical space into three levels: geometric, semantic and decision-making, realizing precise spatial reasoning and action planning for vision-language navigation. These structured representations establish a principled connection between local, partial egocentric observations and global task-level reasoning, while significantly enhancing the agent’s ability to model long-term contextual information, dynamic entity interactions, and physical constraints in complex real-world environments.
Overall, our research constructs a unified technical route of structured granular representation + uncertainty knowledge processing for egocentric visual reasoning, and validates the effectiveness of our methods on multiple self-built and public benchmarks. These works advance the development of more robust, generalizable and real-time embodied AI systems, and provide valuable theoretical insights and engineering references for the integration of rough set, granular computing and egocentric computer vision, laying a solid foundation for the practical application of embodied AI in smart wearables, service robots, medical navigation and other core domains.

Didier Dubois has been an Emeritus Research Advisor at IRIT, the Computer Science Department of Paul Sabatier University in Toulouse, France since 2019. He has been a researcher in the French National Centre for Scientific Research (CNRS) since 1984.
He holds a Doctorate in Engineering from ENSAE, Toulouse (1977), a Doctorat d’Etat from Grenoble University (1983). He is the co-author, with Henri Prade, of two books on fuzzy sets and possibility theory; with Inès Couso and Luciano Sanchez, of a small book on fuzzy random variables, and an editor of 15 volumes on uncertain reasoning and fuzzy sets. Also with Henri Prade, he coordinated the HANDBOOK of FUZZY SETS series published by Kluwer (7 volumes, 1998-2000) including the book « Fundamentals of Fuzzy Sets » which he co-edited. He has contributed more than 200 technical journal papers on uncertainty theories and applications.
He has been a co-Editor-in -Chief of the journal Fuzzy Sets and Systems from 1999 to 2021, an Advisory Editor of the IEEE Transactions on Fuzzy Systems, and a member of the Editorial Board of several technical journals, such as the International Journals on Approximate Reasoning, General Systems, Applied Logic, and Information Sciences among others.
He is a former president of the International Fuzzy Systems Association (1995-1997). He received the 2002 Pioneer Award of the IEEE Neural Network Society, and the 2005 IEEE TFS Outstanding Paper Award. He received the EUSFLAT Scientific Excellence Award in 2012. He is also a recipient of Honorary Doctorates from the Faculté Polytechnique de Mons, Belgium (1997), and Obuda University (Hungary, 2016)

Title: Incomplete information, three-valued logics and formal concept analysis

Abstract: The abstract is available in the following document: Dubois_Didier_IJCRS26_Keynote.pdf

Francisco Herrera received his M.Sc. in Mathematics in 1988 and Ph.D. in Mathematics in 1991, both from the University of Granada, Spain. He is a Professor in the Department of Computer Science and Artificial Intelligence at the University of Granada and Director of the Andalusian Research Institute in Data Science and Computational Intelligence (DaSCI). He’s an academician in the Royal Academy of Engineering (Spain).
He has been the supervisor over 70 Ph.D. students. He has published more than 700 journal papers, receiving more than 187000 citations (Scholar Google, H-index 195). He has been nominated as a Highly Cited Researcher (in the fields of Computer Science 2014 to present, Clarivate Analytics). He acts as editorial member of a dozen of journals.
His current research interests include among others, computational intelligence, data science, trustworthy artificial intelligence, generative AI, general purpose artificial intelligence.

Title: AI Safety through Fuzzy-Rough Boundary Modeling: Monitoring and Anomaly Detection in Open-World Systems

Abstract: Artificial Intelligence systems increasingly operate in open-world environments where unknown classes, distributional shift, and anomalous inputs are unavoidable. Most learning models, however, are developed under closed-world assumptions, enforcing exhaustive decisions even when inputs lie outside validated knowledge. In AI safety-critical contexts, this structural overconfidence becomes a source of risk.
From a rough-set perspective, AI safety can be interpreted as a boundary-recognition problem. Lower approximations represent regions of reliable knowledge, while boundary regions capture epistemic uncertainty and partial evidence. When inputs fall within these boundary areas, principled abstention or escalation mechanisms become necessary. Importantly, rough-set approximations provide interpretable structural explanations, explicitly indicating whether a decision is supported by consistent evidence or lies within a region of ambiguity.
We frame AI safety as a monitoring problem and analyze a fuzzy-rough boundary modeling approach in which rough approximation’s structure epistemic regions and fuzzy compatibility degrees provide graded conformance signals for runtime inspection. These mechanisms generate transparent safety indicators that explain acceptance, rejection, or escalation decisions. As an illustrative example, an image classifier encountering unseen categories can detect boundary proximity and trigger anomaly-aware safety responses instead of forcing classification.
This work positions rough-set boundary modeling as a natural, interpretable, and explainable foundation for monitoring-driven AI safety.