Keynote Speakers

---

Keynote Speakers

Prof. Shyi-Ming Chen (IEEE Fellow, IET Fellow, IFSA Fellow, AAIS Fellow, IETI Distinguished Fellow)
Member of the National Academy of Artificial Intelligence (NAAI),
Fellow of the Pakistan Academy of Engineering (PAE),
Asia University, Taiwan
 

Biography: Shyi-Ming Chen is a Chair Professor in the Department of Computer Science and Information Engineering, Asia University, Taichung, Taiwan. He received the Ph.D. degree in Electrical Engineering from National Taiwan University, Taipei, Taiwan, in June 1991. He is an IEEE Fellow, an IET Fellow, an IFSA Fellow, an AAIS Fellow, an IETI Distinguished Fellow, a Member of the National Academy of Artificial Intelligence (NAAI), and a Fellow of the Pakistan Academy of Engineering (PAE). He was a Chair Professor in the Department of Computer Science and Information Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan. He was the Dean of the College of Electrical Engineering and Computer Science, Jinwen University of Science and Technology, New Taipei City, Taiwan. He was the Vice President of the National Taichung University of Education, Taichung, Taiwan. He was the President of the Taiwanese Association for Artificial Intelligence (TAAI). He was the President of the Taiwanese Association for Consumer Electronics (TACE). He has published more than 600 papers in referred journals, conference proceedings and book chapters. His research interests include Fuzzy Systems, Intelligent Systems, Fuzzy Decision Making, Computational Intelligence, Knowledge-Based Systems, Machine Learning, Deep Learning, Data Mining, Big Data Analysis, Genetic Algorithms, and Particle Swam Optimization Techniques.
He is an Associate Editor of IEEE Transactions on Fuzzy Systems, an Associate Editor of IEEE Transactions on Cybernetics (2019-2024), an Associate Editor of the IEEE Transactions on Systems, Man, and Cybernetics: Systems, an Associate Editor of IEEE Transactions on Artificial Intelligence, an Associate Editor of Knowledge-Based Systems, an Associate Editor of Expert Systems with Applications, an Associate Editor of Information Fusion, an Associate Editor of International Journal on Artificial Intelligence Tools, an Associate Editor of International Journal of Pattern Recognition and Artificial Intelligence, an Associate Editor of International Journal of Fuzzy Systems (2003-2024), an Associate Editor of Journal of Information Science and Engineering, an Associate Editor of Fuzzy Optimization and Decision Making (2016-2024), an Associate Editor of Knowledge and Information Systems (2016-2025), an Editor of International Journal of Intelligent Systems, an Editor of Engineering Applications of Artificial Intelligence (2022-2025), an Editor of Applied Computational Intelligence and Soft Computing, and an Associate Editor of International Journal of Computational Intelligence and Applications.

Speech Title: Fuzzy Forecasting Based on High-Order Fuzzy Time Series and Genetic Algorithms

Abstract: In our daily life, we often use forecasting techniques to predict the weather, the earthquakes, the stock, the temperature, .., etc. Many methods have been presented to deal with forecasting problems. The drawbacks of the traditional forecasting methods are that they cannot deal with forecasting problems whose historical data are linguistic values and their forecasting accuracy rates are not good enough. In this talk, we will present a method for temperature prediction and TAIFEX forecasting based on two-factors high-order fuzzy time series and genetic algorithms. The proposed method gets higher forecasting accuracy rates than the ones obtained by the existing methods. We also will point out some future research directions in this talk.
 

Prof. Hiroki Matsutani
Keio University, Japan

Biography: Hiroki Matsutani received the BA, ME, and PhD degrees from Keio University, Yokohama, Japan, in 2004, 2006, and 2008, respectively. He is currently a Professor in the Department of Information and Computer Science at Keio University. His research interests include computer architecture, interconnection networks, hardware accelerators, and machine learning algorithms.

Speech Title: On-Device Learning for Edge AI: From Algorithms to Practical Applications

Speech Abstract: We are working on on-device learning technologies that enable AI models to train and adapt directly on edge devices with limited computational resources. This approach to on-device learning is highly effective when there is a discrepancy between the training data available beforehand and the data actually collected in the field. It is characterized by its ability to update models on-site to adapt to environmental changes and individual differences. In recent years, AI chips equipped with on-device learning capabilities have emerged, and their commercialization and mass production are steadily progressing. In this presentation, we will introduce on-device learning, covering everything from its underlying algorithms to its practical applications.

 

Prof. Zheng Yan (IEEE Fellow, IET Fellow, AAIA Fellow, and AIIA Fellow)
Xidian University, China

Biography: Dr. Zheng Yan is currently a Distinguished Professor at Xidian University, China. She earned the Doctor of Science in Technology from Helsinki University of Technology. She is a Stanford World top 2% scientist, an Elsevier highly cited Chinese researcher, and a ScholarGPS World Top 0.05% Highly Ranked Scholar. Her research interests are in trust, security, privacy, and data analysis. She has published more than 470 papers in prestigious journals and conferences, with 300+ as the first or corresponding author. She has authored two English books, used for teaching for nearly a decade. She invented 220+ patents including 50 PCT patents, with more than 150 patents adopted by industry, most of them are solely invented by her. Some of these patents have entered international standards and widely used. She has received numerous awards, including the Nokia Distinguished Inventor, IEEE TCSC Award for Excellence, IEEE HITC Industrial Impact Award, IEEE TEMS Distinguished Leadership, N²Women Star in Computer Networking and Communications, three EU awards, two IEEE TC best journal paper awards, Shaanxi Natural Science Award, etc. She is currently a member of IEEE Fellow Committee. She serves as a Co-EiC of Information Sciences and an Area Editor/Associate Editor for 60+ esteemed journals. She founded the IEEE International Conference on Blockchain and serves as its Steering Committee Co-chair. She has contributed to more than 50 conferences as a General Chair or TPC Chair and delivered more than 50 keynotes and invited talks. She is a Member of Finnish Academy of Science and Letters, and a Fellow of IEEE, IET, AAIA, and AIIA.

 

Speech Title: Decentralized Trust Management with Privacy Preservation

Abstract: Blockchain provides a decentralized, tamper-resistant, and transparent ledger that enables secure, trustworthy, and auditable data sharing among mutually untrusted parties without relying on a central authority. As one of the most influential decentralized technologies, blockchain has shown tremendous potential across a wide range of applications, with decentralized trust management being a particularly promising area. However, its inherent transparency and openness also introduce significant privacy challenges, making privacy preservation a fundamental research issue. Although substantial progress has been made, many open problems remain, especially in achieving privacy-preserving decentralized trust management.
In this talk, I will present our recent research on blockchain-based decentralized trust management with privacy preservation in two representative application scenarios: cross-chain transactions and integrated heterogeneous networks. I will introduce the key techniques we have developed to address these challenges and demonstrate their effectiveness through proofs, experiments and several prototype systems. Some of these prototypes have already been transferred to a leading telecommunications company for further development and practical deployment.

 

Invited Speaker

---

Prof. Seungshik Kang ,Kookmin University, Republic of Korea

Biography: He received B.S., M.S., and Ph.D. degrees in Computer Science from Seoul National University in 1986, 1988 and 1993, respectively. He is working for Kookmin University as a full professor. His research interests include natural language processing, text mining, big data processing, and machine learning.
His doctoral dissertation focused on Korean morphological analysis, and the new algorithm he proposed laid the foundation for the development of a practical Korean morphological analyzer. He has made significant contributions to the field of Korean information processing, and the morphological analyzer he developed has been widely used as a Hangul indexer in domestic and international search engines.
Recently, his interest in deep learning NLP is Hangul tokenization and embedding techniques. He is studying the impact of tokenizer performance on the learning speed and accuracy of deep learning NLP systems.

 

Speech Title: Recent Research Issues in Natural Language Processing and Large Language Models

Abstract: Recently, the field of natural language processing has been undergoing rapid changes due to the advancement of transformer-based large language models. Transformers utilize a self-attention mechanism to simultaneously learn the relationships between all words within a sentence, serving as the foundational technology for subsequent BERT and GPT models. In particular, GPT models have significantly improved natural language generation capabilities through large-scale data and parameter expansion, enabling them to perform various tasks such as human-like text generation, question answering, translation, and code writing. However, the advancement of LLMs signifies more than just simple performance improvements. Issues such as increased computational costs due to larger model sizes, bias in training data, hallucination, and a lack of inference ability continue to be raised. While early LLM research focused on expanding model size and increasing pre-training data, recent research directions have expanded to include enhancing inference capabilities, efficient learning methods, retrieval-augmented generation, multimodal processing, agent systems, and input representation optimization. However, LLM still faces limitations such as hallucinations, high computational costs, data quality issues, and differences in expression across languages, and recent research is shifting in the following directions. First, research is focusing on efficient learning and inference methods rather than increasing model size. Second, the direction of research is shifting from simple language generation to the development of models equipped with complex reasoning capabilities. Third, research on agent-based systems capable of external knowledge retrieval and tool utilization is expanding. Fourth, research on tokenization and input representation to improve language expression itself is becoming increasingly important. This study systematically analyzes these latest trends in NLP research and suggests research directions that should be considered specifically in the Korean language environment. In particular, in non-English languages like Korean, linguistic characteristics such as morpheme structure, conjugations of particles and endings, compound nouns, and spacing variations have a significant impact on model performance. This study analyzes major recent research trends in the fields of NLP and LLM, and examines inference-centric models, efficient learning, knowledge retrieval-based generation, multimodal models, semantic-based tokenization, and research directions specific to the Korean language. Furthermore, it is suggested that optimizing data quality and language representation methods will become important research tasks in future NLP research, rather than simply expanding model size.