Pablo A. Estévez received his professional electrical engineering (EE) title from Universidad de Chile in 1981 and the M.Sc. and Dr. Eng. degrees from the University of Tokyo, Japan, in 1992 and 1995, respectively. He is a Full Professor in the Electrical Engineering Department at the University of Chile and former Chairman of the Electrical Engineering Department from the period 2006-to 2010. Prof. Estévez is an IEEE Fellow. He served as President of the IEEE Computational Intelligence Society (CIS) for the term 2016-2017, Vice-president of Finances (2019-2024), Vice-president of Members Activities (2011-2014), CIS ADCOM Member-at-Large (2008-2010), and CIS Distinguished Lecturer (2006-2011). He has served as an Associate Editor of the IEEE Transactions on Neural Networks (2007-2012), and as Associate Editor of the IEEE Transactions on Artificial Intelligence (2020-2024). Prof. Estévez is the recipient of the 2019 IEEE CIS Meritorious Service Award and the 2019 IEEE Latin-America Eminent Engineer Award.  
Prof. Estévez served as conference chair of the International Joint Conference on Neural Networks (IJCNN), held in July 2016, in Vancouver, Canada, and general co-chair of the IEEE World Congress on Computational Intelligence, IEEE WCCI 2018, held in Rio de Janeiro, Brazil, in July 2018. Prof. Estévez has been a keynote speaker at GAITC 2023 (China), IEEE ANDESCON (2022), CBIC 2019 (Brazil), Chilecon 2019 (Chile), WSOM 2016 (USA), IJCRS 2016 (Chile), LA-CCI 2016  (Colombia), LA-CCI 2015 (Brazil), IEEE SSCI 2014 (USA), AISC 2008 (Poland), among other conferences. Prof. Estévez served as PI of the Conicyt-NSF international collaboration project on Big Data in Astronomy (2015-2018, US$300,000), carried out together with the University of Washington. He is a founder and co-PI of the Millennium Institute on Astrophysics (2014-2024, US$10,000,000), and co-Pi of the Millennium Institute for Intelligent Healthcare Engineering-IHEALTH (2023-2032). Prof. Estévez collaborates with researchers from Harvard University, John Hopkins University, University of Washington, University of Florida, Pantheon-Sorbonne, and Nancy University. Prof. Estévez has been an Invited Researcher with the NTT Communication Science Laboratory, Kyoto, Japan; the Ecole Normale Supérieure, Lyon, France; and a Visiting Professor at the Pantheon-Sorbonne University, Paris, France; and the University of Tokyo, Tokyo, Japan. Prof. Estévez has co-authored more than 150 articles in journals and conferences (Google Scholar h-index = 36). His research interests include artificial intelligence, big data, deep learning, artificial neural networks, self-organizing maps, data visualization, feature selection, information theoretic learning, time series analysis, and advanced signal and image processing, with applications to astronomy, energy and medicine.

Empowering Astronomy and Medicine through Transformers and LLMs

Pablo A. Estévez,
Electrical Engineering Department
Universidad de Chile

Abstract: Deep learning attention-based models (transformers) are the new technology underlying the current AI revolution, particularly in natural language processing through large language models (LLMs). In this talk, we will introduce transformers and LLMs. We apply transformers to process astronomical time series to classify supernovae events and compare results with state-of-the-art machine learning models such as recurrent neural networks and random forests. Then, we introduce ATAT, a multimodal transformer that combines time series and tabular data. ATAT is applied to the synthetic dataset from the ELAsTICC challenge, reaching a macro F1-score of 82.9±0.4 using 20 classes. We also apply LLMs to the text-to-SQL parsing problem in astronomical datasets. Next, we apply transformers to detect sleep spindles in electroencephalograms (EEGs). Finally, we use multimodal LLMs for generating text reports from brain medical images.

 James M. Keller is a University of Missouri Curators Distinguished Professor Emeritus in the Electrical Engineering and Computer Science Department on the Columbia campus. His research interests center on computational intelligence with a focus on problems in computer vision, pattern recognition, and information fusion including bioinformatics, spatial reasoning, geospatial intelligence, landmine detection and technology for eldercare.  Professor Keller has been funded by a variety of government and industry organizations and has coauthored well over 500 technical publications.  
Jim is a Life Fellow of the IEEE, is an IFSA Fellow, and a past President of NAFIPS. He received the 2007 Fuzzy Systems Pioneer Award and the 2010 Meritorious Service Award from the IEEE Computational Intelligence Society. He won the 2021 IEEE Frank Rosenblatt Technical Field Award. Jim was Editor-in-Chief of the IEEE Transactions on Fuzzy Systems (1999 – 2004), followed by being the Vice President for Publications of the IEEE Computational Intelligence Society from 2005-2008, then as an elected CIS Adcom member, and finished another term as VP Pubs (2017-2020). He was President of IEEE CIS for 2022 – 2023. and has served as the IEEE TAB Transactions Chair and as a member of the IEEE Publication Review and Advisory Committee from 2010 to 2017. He was Vice Chair of the IEEE Publication Services and Products Board, chaired an AdHoc Committee on AI in the Publications Domain, and now serves as the chair of the strategic planning committee of that board. He has had many conference positions and duties over the years.

Checking on the Stream

Jim Keller*
Electrical Engineering and Computer Science
University of Missouri

Abstract: With the explosion of ubiquitous continuous sensing (something Lotfi Zadeh predicted as one of the pillars of Recognition Technology in the late 1990s), on-line streaming clustering is attracting more and more attention.  I was drawn into this world mainly due to our desire to continuously monitor the activities, and health conditions, of older adults in a large interdisciplinary eldercare research group. Roughly, the requirements are that the streaming clustering algorithm recognize and adapt clusters as the data evolves, that anomalies are detected, and that new clusters are automatically formed as incoming data dictate. An advantage of a streaming model is that data trends can be examined as they occur, and alerts could be generated as feature vectors representing activity move “toward” cluster boundaries. 
Streaming clustering is fundamentally different from static traditional clustering where the most successful models use some form of alternating optimization in an iterative exploration of a complete dataset.  When the clustering algorithm finishes, there is a variety of so-called cluster validity indices to estimate the goodness of the final partition, be it crisp, probabilistic, fuzzy, or possibilistic. This is not possible in the streaming mode, since data arrives in a continuous fashion and decisions need to be made instantaneously.  In most cases, the data is then discarded and only cluster “footprints” are retained. Is the situation then hopeless?  This would be a short talk if so!
In this talk, I will present my thoughts on streaming clustering in general, looking in a little more detail at the MU streaming clustering algorithm.  I will briefly show an early warning approach of health changes in eldercare monitoring based on monitoring sensor streams.  The main goal, however, is to dig into the streaming equivalent of static cluster validity measures that we call incremental streaming indices.  I will discuss several such indices including two new fuzzy set based measures of incremental cluster growth and drift. 

* research done with lots of friends, almost too numerous to be named!

Bernadette Bouchon-Meunier is a director of research emeritus at the National Centre for Scientific Research and Sorbonne University, the former head of the department of Databases and Machine Learning in the Computer Science Laboratory of the University Pierre et Marie Curie-Paris 6 (LIP6). She supervised 52 PhD students. She is the Editor-in-Chief of the International Journal of Uncertainty, Fuzziness and Knowledge-based Systems and the Co-executive director of the IPMU International Conference held every other year since 1986. B. Bouchon-Meunier is the (co)-editor of 30 books, and the (co)-author of six.
She has (co)-authored more than 450 papers on approximate and similarity-based reasoning, as well as the application of fuzzy logic and machine learning techniques to decision-making, data mining, risk forecasting, information retrieval, user modelling, sensorial and emotional information processing.
She was elected President of the IEEE Computational Intelligence Society for 2020-2021. She is an IEEE Life Fellow, an International Fuzzy Systems Association Fellow and an Honorary Member of the EUSFLAT Society. She received the 2012 IEEE Computational Intelligence Society Meritorious Service Award, the 2017 EUSFLAT Scientific Excellence Award, the 2018 IEEE CIS Fuzzy Systems Pioneer Award, the 2019 Outstanding Volunteer Award of the IEEE France Section and the 2024 IEEE Frank Rosenblatt award

Fuzzy Approaches to Explainable Artificial Intelligence

Bernadette Bouchon-Meunier
National Centre for Scientific Research
Sorbonne University

Abstract: Explainable Artificial Intelligence has been at the core of many developments in Artificial Intelligence in the recent years, following the DARPA incentives (https://www.darpa.mil/program/explainable-artificial-intelligence) to both produce more explainable artificial intelligence models while maintaining good learning performances, and enable the user to easily interact with intelligent systems. Several related concepts are inherent in the quality of intelligent systems, such as their understandability, their expressiveness or their interpretability. The latter has been in particular extensively investigated in the construction and analysis of fuzzy intelligent systems. We will review several of these aspects, mainly dealing with the capacity of an intelligent system to explain how it obtains results, and to provide the user with easily understandable outcomes, in light of the fuzzy paradigm. We will show that fuzzy models participate effectively in solutions to achieve one or the other of these goals, going beyond classic fuzzy rule-based systems.