Abstract: This tutorial provides important soft skills that are important for successful careers: skills beyond technical and scientific excellence. We address topics on writing research grant proposals, finding appropriate funding resources, research ethics, conflicts of interest, etc. We also discuss writing good papers, presenting/communicating novel ideas clearly, learning citation, copyright and plagiarism rules/customs, finding/choosing the right journals to submit, peer reviewing, serving on conference review boards and journal editorial boards, generally developing a sense of belonging to the community. Other essential skills include being a mentor and a mentee, sharing knowledge with others and overall, benefitting from professional networking. Target audience of the tutorial is young professionals and others who might be interested in learning about these (often overlooked) aspects of technical careers. The interactive format with audience Q & A will encourage open discussion and provide useful soft skills that are important for professional success.

Organizers: Fahmida N. Chowdhury, Basabdatta Sen Bhattacharya, Sushmita Mitra

TU4 webpage: https://www.binnlabs-goa.in/whats-new/tutorialieee_fuzz_2025

Abstract: The goal of this tutorial is to make participants familiar with fuzzy rule based networks and their inherent interpretability which makes them suitable for building explainable artificial intelligence models. As opposed to current machine learning models which can reflect mainly statistical correlations between input and output variables, fuzzy rule based network models use machine reasoning that allows them to reflect also causal relationships by means of intermediate variables in their network structure. The nodes of fuzzy rule based networks are fuzzy systems represented by rule bases and the connections between these nodes are outputs from and inputs to these rule bases. In this context, apart from being a structural counterpart for a neural network, a fuzzy rule based network is also a conceptual generalisation of a fuzzy system and a bridge between two established types of fuzzy systems- flat and a hierarchical.

Organizers: Alexander Gegov, Farzad Arabikhan, Alexandar Ichtev

TU5 webpage: https://sites.google.com/port.ac.uk/tutorial/home?pli=1

Abstract: This tutorial is limited to hard k-means and some of its soft relatives. It begins with a short history of Legendre’s method of least squares, which leads to two very different algorithms commonly called k-means. Then the algebraic and geometric structure of partition sets underlying all hard, probabilistic, possibilistic and fuzzy clustering algorithms is presented. The structural theory illuminates some little-known facts about k-means and some of its soft relatives. For example, it is often stated that hard k-means has terminated at a local minimum of its objective function; the structural theory shows this to be impossible. Convex decomposition of soft partitions is discussed. Finally, selected soft generalizations of k-means that should be of interest to practitioners in this area are briefly discussed.

Organizers: James C. Bezdek

TU6 webpage: NA

Abstract: Conformal prediction is a rising topic of interest for performing uncertainty quantifi- cation (UQ) in machine learning procedures, and more generally to perform UQ from a statistical population.
The goal of the proposed tutorial is :
• to introduce the audience to the basic ideas of conformal prediction, focusing on its key principles, fundamental concepts and ways to apply them in classification and regression problems;
• to provide some details on some recent developments of conformal prediction that could be of interest to the FUZZ’IEEE audience, such as its application to multi- target problems or its use in weakly supervised learning.
While theoretical challenging questions will be mentioned during the tutorial, we will not go deep in this direction, and will mainly give pointers to the more theoretically inclined audience.

Organizers: Sébastien Destercke, Soundouss Messoudi, Bruce Cyusa Mukama

TU7 webpage: https://fuzzieee2025.conf.lip6.fr/wp-content/uploads/2025/01/Fuzz_ieee_2025_TU7_Conformal_prediction.pdf

Panellists: Jonathan Garibaldi (Chair), Keeley Crockett, Alexander Gegov, Uzay Kaymak, Joao Sousa, Hussein Abbass

Abstract: The panel members will argue that the artificial intelligence (AI) community needs to escape the trap of explainable artificial intelligence (xAI) by growing more research on understandable artificial intelligence (uAI). They will provocatively term xAI a trap because it has caused some AI researchers to see it as the “aim” rather than a “means” They will discuss why uAI is a better way forward and present a framework for uAI to define research directions that go beyond xAI.

Understandability is one of the key components required for regular citizens to build trust in AI in their everyday lives. But what does understandability actually look like for people with different lived experiences, in different application domains? The panel will explore the role of interdisciplinary research and participatory AI approaches in capturing what users really want. It will also discuss the ability of fuzzy systems to improve understanding for general users due to the inherent interpretability of the fuzzy rules and the associated linguistic terms.

Recent advances in machine learning have led to a shift in the perception of AI from being a field in academic labs to a commodity ready for end users to create wealth. This revolution caused a spike in the need for users to understand AI, including AI models, develop confidence in their output, and trust in adopting them, especially in safety critical applications. DARPA sensed this signal as early as 2016 and established a research program in xAI.

DARPA’s promotion of xAI contributed to an exponential growth in the number of papers on the topic. In most of these papers, the generation of explanations and their integration within the underlying machine learning models was quite successful as far as expert users are concerned. However, little advancement was made on “understanding” of explanations by general users who usually do not have a sufficient level of domain knowledge.

On the basis of the considerations above, panel members will argue that significant research opportunities exist for uAI, in which: 1) explanation is only one type of evidence needed for understanding and other forms of evidence, such as statistical evidence and demonstrations through simulation and digital twins, need more extensive research; and 2) explanation methods for machine learning models will need to be designed with “understanding” in mind.