 | Prof. Nikhil R. Pal, Indian Statistical Institute, IndianIEEE Fellow, Fellow of the Indian National Science Academy Nikhil R. Pal is an INSA Senior Scientist in the Electronics and Communication Sciences Unit of the Indian Statistical Institute (ISI). He is an Honorary Visiting Professor of the South Asian University, India. He was a former professor of ISI and was the Head of the Center for Artificial Intelligence and Machine Learning. He served as an INAE Chair Professor at ISI as well as a Chair Professor at the National Chiao Tung University, Taiwan. He also served as a visiting Professor of China University of Petroleum, East China; Huazhong University of Science and Technology, Wuhan; and the City University of Hong Kong. His current research interest includes brain science, computational intelligence, machine learning and artificial intelligence. He was the Editor-in-Chief of the IEEE Transactions on Fuzzy Systems for the period January 2005-December 2010. He served/has been serving on the editorial /advisory board/ steering committee of several journals including the International Journal of Approximate Reasoning, Applied Soft Computing, International Journal of Neural Systems, Fuzzy Sets and Systems, IEEE Transactions on Fuzzy Systems and the IEEE Transactions on Cybernetics. He is a recipient of the 2015 IEEE Computational Intelligence Society (CIS) Fuzzy Systems Pioneer Award and 2021 IEEE CIS Meritorious Service Award. He has given many plenary/keynote speeches in different premier international conferences in the area of computational intelligence. He has served as the General Chair, Program Chair, and co-Program chair of several conferences. He was a Distinguished Lecturer of the IEEE CIS (2010-2012, 2016-2018, 2022-2024) and was a member of the Administrative Committee of the IEEE CIS (2010-2012). He has served as the Vice-President for Publications of the IEEE CIS (2013-2016) and the President of the IEEE CIS (2018-2019). He is a Fellow of the West Bengal Academy of Science and Technology, Institution of Electronics and Tele Communication Engineers, National Academy of Sciences, India, Indian National Academy of Engineering, Indian National Science Academy, International Fuzzy Systems Association (IFSA), The World Academy of Sciences, and a Fellow of the IEEE, USA. (www.isical.ac.in/~nikhil). Title: In the Age of Trillion Parameters, Does Dimensionality Reduction Still Matter? Abstract: We are living in the era of Large Language Models (LLMs) and Artificial Intelligence (AI)—a time when AI systems, particularly LLMs, are advancing at an unprecedented pace. Much of this progress has been driven by the philosophy that “bigger is better”: larger datasets, deeper architectures, and increasingly massive models—some now exceeding trillion parameters. This naturally raises an important question: Is the reduction of model complexity or dimensionality still relevant today? In my view, the answer is “YES”. Not every problem is a “big data” problem, nor are all tasks text- or image-based. Moreover, many real-world challenges, such as identifying disease biomarkers or discovering target molecules in drug design, cannot rely solely on black-box models trained on uncurated data. This does not diminish the value of LLMs in such domains; rather, it reinforces the importance of dimensionality control for achieving reliable and interpretable results. Even within LLMs, dimensionality reduction (DR) plays a crucial role: embedding mechanisms transform high-dimensional data into compact, informative feature spaces. In this talk, first we shall show an example on how DR can help LLMs and also discuss some important theoretical results justifying DR. Then we shall briefly discuss how model complexity can be reduced, in general, and the important role played by feature/sensor selection for this. Using feed-forward neural networks, we will demonstrate how concepts like feature attenuators and group Lasso facilitate DR via sensor (as well as feature) selection while maintaining control over the relevance and redundancy of the selected sensors (features). We shall also demonstrate how very useful fuzzy rule-based systems can be designed in an unsupervised manner for data visualization through manifold learning. Ultimately, this talk underscores that the timeless wisdom of great scientists remains profoundly relevant in the age of AI and LLMs: “Make everything as simple as possible, but not simpler.” — Albert Einstein (1879–1955) “Nature is pleased with simplicity. And nature is no dummy.” — Isaac Newton (1643–1727) |
 | Prof. Hani Hagras, University of Essex, UKIEEE Fellow Hani Hagras is a Professor of Artificial Intelligence (AI), Director of Impact, Head of the Artificial Intelligence Research Group, and Director of the Computational Intelligence Centre, University of Essex UK. He holds eleven international industrial patents and has published over 500 technical papers. He is also the co-author and/or co-editor of five books. His current research interests include Explainable Artificial Intelligence (XAI), Generative AI, and computational intelligence, with applications in finance, life sciences, uncertainty management, Internet of Things, cyber-physical systems, intelligent robotics, and the intelligent control of industrial processes. He is the recipient of 2026 IEEE Fuzzy Systems Pioneer Award. He is a Fellow IEEE, Fellow of the European Academy of Sciences, Fellow of the Institution of Engineering and Technology (IET), Principal Fellow of the UK Higher Education Academy (PFHEA), Fellow of the Asia-Pacific Artificial Intelligence Association (AAIA) and Fellow of the Artificial Intelligence Industry Alliance (AIIA). He is ranked among the top 2% of the most-cited scientists in the world (Scopus, August 2024), and is recognized as a Highly Ranked Scholar by ScholarGPS, placing him in the top 0.05% of all scholars worldwide. His work has been supported by major research councils and industry. He has received numerous prestigious international awards where he was awarded by the IEEE CIS, the 2026 Fuzzy Pioneer Award, the 2004 and 2010 Outstanding Paper Award in the IEEE Transactions on Fuzzy Systems, the 2015 and 2017 Global Telecommunications Business Awards (with British Telecom), Named Distinguished Lecturer by the IEEE Computational Intelligence Society (2016), Multiple Best Paper Awards at leading international conferences, including, 2006 and 2014 IEEE International Conferences on Fuzzy Systems, 2016 BCS SGAI International Conference on Artificial Intelligence and others. He acted as the Chair of the IEEE CIS Chapter, which won 2011 IEEE CIS Outstanding Chapter Award. He was also the recipient of the IET Knowledge Networks Award. Prof. Hagras has served as a long-standing Associate Editor for several leading journals, including IEEE Transactions on Fuzzy Systems, IEEE Transactions on Artificial Intelligence, Knowledge-Based Systems, Cognitive Computing, among others. He has also chaired numerous international conferences, notably as General Co-Chair of the 2007 IEEE International Conference on Fuzzy Systems, Programme Chair for the 2017 and 2021 IEEE International Conference on Fuzzy Systems and others. Title: AI Transparency: The Key to Unlocking User Confidence Abstract: The explosive growth of computing power and the ever-increasing flood of data have reignited global interest in Artificial Intelligence (AI). Yet, as AI systems become more powerful, they also become more complex—and often more mysterious. Many of today’s models operate as “black boxes,” producing impressive results but offering little insight into how those results are achieved. However, recent studies have concluded that without transparency, AI cannot scale responsibly in real world applications and services. Explainability must sit at the core of AI where explainability has moved from “nice-to-have” to mandatory by recent customer, industry and regulators demands. The accuracy-versus-transparency dilemma is no longer an issue with the rise of explainable by design AI which can earn trust, drive adoption, and meet regulatory expectations. |
 | Prof. Peng Ren, China University of Petroleum (East China), ChinaPeng Ren received the BEng and MEng degrees both in electronic engineering from Harbin Institute of Technology, China, and PhD in computer science from the University of York, UK. He is currently a full professor with the College of Oceanography and Space Informatics, China University of Petroleum (East China). He is the director of Shandong Youth Innovative Team of Offshore Unmanned Observation and also the director of Qingdao International Research Center of Intelligent Forecast and Detection of Oceanic Catastrophes. He received the K. M. Scott Prize from the University of York, and the Eduardo Caianiello Best Student Paper Award from 18th International Conference on Image Analysis and Processing as one co-author. He serves as associate editors of IEEE TGRS, IEEE JOE, and PR. His research interests include learning based underwater imaging and remote sensing, and on-orbit FPGA computation, etc. He is an IEEE OES distinguished lecturer. Title: Machine Learning Paradigms from Mutual Guide to Confucius Tri-Learning With Applications to Remote Sensing Abstract: This presentation introduces a series of machine learning paradigms developed from the concept of Mutual Guide to Confucius Tri-Learning, with applications in remote sensing. Beginning with a comparison to Generative Adversarial Networks (GANs), Mutual Guide is proposed as a cooperative framework where models with identical structures but different initializations guide each other to improve performance, particularly under small training data conditions. Further extensions include Cross Supervised Learning for cloud detection and Mutually Beneficial Guide for semi-supervised cloud removal, both leveraging labeled and unlabeled data through mutual supervision and objective guidance. Mutual Feature Guide is also presented for sea ice classification, where guiding features are exchanged between models to enhance representation learning. The core innovation, Confucius Tri-Learning, integrates adversarial learning with mutual guidance, inspired by the Confucian idea of learning from both good and bad examples. This paradigm is applied to SAR target classification, showing improved generalization and robustness. Theoretical analysis suggests that Confucius Tri-Learning achieves a tighter VC bound, offering a principled framework for learning from limited and heterogeneous data. The work highlights a cohesive evolution from adversarial to collaborative learning, with broad applicability in remote sensing tasks. |
 | Prof. Dunwei Gong, Qingdao University of Science and Technology, ChinaIEEE Senior Member Dunwei Gong is a professor at Qingdao University of Science and Technology. He serves as the director of the School of Microelectronics. He was selected as the specialty expert in the Taishan Scholars program, Shandong province, Top 2% highly cited scientists, a fellow of Asia-Pacific Artificial Intelligence Association. His main research areas include intelligent optimization theory and methods, as well as their applications in operation scheduling for integrated energy systems, parameter optimization for integrated circuit devices, industrial production line optimization, and unmanned aerial vehicle inspection. He has hosted several national and provincial-level research projects, including one National Key R&D Plan project, seven National Natural Science Foundation projects (including one major project), and one major basic research project of Shandong province. He has won five significant scientific awards in China, including the Natural Science Award of Higher Education Institutions in China, the Jiangsu Provincial Science and Technology Award for Sciences, the Shandong Provincial Natural Science Award, and the Natural Science Award of the Chinese Automation Society. Title: What have we done in industrial applications of intelligent optimization and learning? Abstract: In recent years, there have been significant research achievements in intelligent optimization and learning. However, these accomplishments have not yet seen widespread application in industrial production. Leveraging my involvement in multiple national and provincial-level research projects, this report will focus on practical applications of intelligent optimization and learning methods within industrial scenarios such as operation scheduling for integrated energy systems, parameter optimization for integrated circuit devices, and embodied intelligence for stacking tools in packaging lines. Additionally, the report will outline future research efforts that aim to advance these technologies further. |