On September 24, Academician Nikhil R. Pal from the Indian Statistical Institute gave an invited academic report titled Several Critical Issues in Computational Intelligence Requiring Urgent Attention to faculty and students of the Joint Laboratory of Cross-Media Big Data. The presentation covered a number of open problems in computational intelligence, with a focus on cutting-edge topics including rule specificity, rule conflict and redundancy in high-dimensional fuzzy systems, dimensionality reduction and feature selection, Matryoshka-style representation learning, and the trustworthiness of computational intelligence systems.
Academician Pal noted that with the rapid advancement of artificial intelligence technologies, fuzzy systems and neural networks have achieved promising performance on complex tasks. Nevertheless, challenges such as insufficient rule specificity, rule conflict and redundancy in high-dimensional settings still limit the interpretability and reliability of such systems. Therefore, computational intelligence systems should pursue not only prediction accuracy but also concise and clear knowledge representation as well as trustworthy decision-making capacity.
The lecture centered on constructing computational intelligence systems that integrate accuracy, interpretability and trustworthiness. Analyzing rule specificity in fuzzy systems, rule conflict and redundancy, dimensionality reduction and feature selection, and Matryoshka nested fuzzy systems, Academician Pal proposed that existing approaches should be improved from the perspectives of rule design, representation learning and training constraints, so that models remain competitive in performance while becoming more compact, transparent and interpretable.
Academician Pal emphasized that these issues carry both theoretical research value and practical application significance. In fields such as medical diagnosis, risk assessment and complex engineering decision-making, intelligent systems are expected to produce accurate outputs while possessing the capability to identify unseen classes, explain reasoning behind judgments and guarantee algorithmic fairness. In-depth research on these open problems will facilitate theoretical innovations in computational intelligence and its dependable deployment in critical scenarios.
After the report, Academician Pal engaged in in-depth discussions with faculty and students present on topics including mathematical characterization of rule conflict and redundancy, and theoretical foundations for feature selection. He encouraged faculty and students to forge closer links between theoretical methodological innovation and engineering applications, and offered targeted and inspiring advice drawing on his own research experience.
Nikhil R. Pal is Fellow of the Indian National Science Academy, the Indian National Academy of Engineering, the National Academy of Sciences, India, and The World Academy of Sciences. He is also Fellow of the International Fuzzy Systems Association and the Institute of Electrical and Electronics Engineers (IEEE). He currently serves as Professor in the Electronics and Communication Sciences Unit at the Indian Statistical Institute and conducts research at the Institute’s Centre for Artificial Intelligence and Machine Learning. His main research interests include brain science, computational intelligence, machine learning and data mining. He was Editor-in-Chief of the IEEE Transactions on Fuzzy Systems (2005–2010), Vice President for Publications of the IEEE Computational Intelligence Society (2013–2016), and President of the IEEE Computational Intelligence Society (2018–2019). He received the IEEE Computational Intelligence Society Fuzzy Systems Pioneer Award in 2015.
[Editor:Ma XiangYi, Zhang YaoGuang]