Prof. Xiangen HU began his academic journey in applied mathematics, earning his Bachelor's and Master's degrees from Huazhong University of Science and Technology in 1982 and 1985, respectively. He then moved to the United States to further his education, obtaining a Master's in social sciences in 1991 and a Ph.D. in cognitive psychology in 1993.
Before his current position as a chair professor in learning sciences and technologies at PolyU, Prof. HU held several positions. He was a professor in the Departments of Psychology, Electrical and Computer Engineering, and Computer Science at The University of Memphis (UofM) for 30 years, where he also worked as a senior researcher at the Institute for Intelligent Systems (IIS). His leadership roles included serving as a professor and Dean of the School of Psychology at Central China Normal University (CCNU), leading the Advanced Distributed Learning (ADL) Partnership Laboratory at UofM, and working as a senior researcher at the Key Laboratory of Adolescent Cyberpsychology and Behavior, backed by the Chinese Ministry of Education.
Prof. HU's research focuses on four key areas: developing mathematical models to decode human cognitive behavior, specializing in research design and statistical analysis particularly for categorical data using general processing tree models, delving into artificial intelligence for knowledge representation, creating computerized tutoring systems, and enhancing distributed learning technologies.
His work has attracted significant funding from prestigious bodies like the US National Science Foundation, the US Institute of Education Sciences, the Advanced Distributed Learning initiative of the US Department of Defense, the US Army Medical Research Acquisition Activity, the US Army Research Laboratories, and the US Office of Naval Research. As the lead principal investigator, Prof. HU has managed projects with over $10 million in funding, and as a co-principal investigator, he has been involved in projects amassing more than $30 million in grants.
Generative AI can produce increasingly sophisticated answers, but educational design requires more than answer generation. How can educators translate principles from the learning sciences into purposeful, interactive learning experiences?
This keynote explores how AI can make educational ideas operational through working technology demonstrations. KnowledgeExplorer illustrates approaches to structuring exploration and questioning; AlwaysAI demonstrates the design of purposeful learning activities and AI interactions; and SafeguardAI brings attention to the boundaries and safeguards needed in educational settings. Together, these examples connect learning principles with concrete design choices about learner participation, instructional support, and teacher oversight.
The presentation follows a common sequence: identify a learning principle, demonstrate an interaction that puts it into practice, and formulate a research question for subsequent evaluation. The demonstrations illustrate technological capabilities and instructional possibilities rather than establish learning effectiveness. They provide a basis for discussing what evidence would be needed to assess independent performance, retention, transfer, and responsible use.
The central argument is that AI’s educational potential extends beyond generating content: it gives educators new ways to build, examine, and refine learning experiences grounded in learning sciences.
Please let me know if there is a required abstract length or format. Could you also confirm the allotted time for the keynote, including questions, so I can plan the demonstrations accordingly?
Professor Yanjie Song is Professor in the Department of Mathematics and Information Technology, Associate Co‑Director of AAPSEF, Associate Director of CELT, and Assistant Dean of the Graduate School at The Education University of Hong Kong. Her research interests focuses on AI and the metaverse in education, technology‑enhanced language learning, and multimodal learning analytics. She has pioneered a suite of award‑winning educational technologies in different international innovation invention competitions, including VocabGo, Learningverse, and EmoCare, as well as new platforms LearningverseVR, and Embodyverse, with several patents granted to her team. She has secured substantial competitive funding, holds leadership roles in APSCE, and is listed among the world’s top 2% most‑cited scientists by Stanford University.
In this talk, I will discuss how applications of generative and embodied AI are transforming education through immersive learning environments. I will introduce five platforms developed by our research team at The Education University of Hong Kong and their applications in education briefly. Web-based Learningverse is a 3D metaverse platform that uses AI-driven digital humans to support interactive and collaborative learning. LearningverseVR extends these functions into virtual reality, where intelligent agents enhance engagement and provide personalised guidance. Embodyverse integrates embodied AI robots into mixed reality to enrich hands-on science learning. iChat-tutor is an AI-powered digital human chat platform that offers customisable interactive learning experiences for teachers and students of all ages. Drawing on empirical examples, the talk highlights how these platforms have the potential to enhance learner engagement, improve performance, and enable new forms of personalised, collaborative, and immersive learning for the future of education.
Peter Reimann is a professor in the Sydney School of Education and Social Work His research focuses on cognitive learning, educational computing, e-learning, multimedia and knowledge-based learning environments, and methods for evaluating technology-supported education. His current work examines individual and collaborative problem-solving, technology-supported learning, and semantic technologies for education. Peter has extensive experience in teaching, research leadership, and the design and evaluation of educational technologies, including hypermedia systems, simulations, and intelligent tutoring systems. He has contributed to numerous European Commission-funded research projects, worked as a reviewer for European IT research and development programs, and founded a Germany-based e-learning consultancy. From 2003 to 2010, he co-directed the University of Sydney’s CoCo Research Centre and, from 2005 to 2010, led the Faculty’s ICT unit. He's currently editor-in-chief of the International Journal of Computer-supported Collaborative Learning (IJCSCL).
Inquiry-based learning has become a cornerstone of contemporary science education. Yet many students still struggle to connect classroom investigations with the work of scientists, engineers, and other STEM professionals. This presentation explores simulated internships, or Virtual Internship Simulators (VIS), as an emerging pedagogical approach that embeds scientific inquiry within authentic professional practice. Rather than asking students merely to think like scientists, this approach invites them to engage in the roles, tasks, and reasoning of scientific work.
In this presentation, I outline an emerging pedagogy of simulated internships for science and engineering education. Drawing on research on cognitive apprenticeship, challenge-based learning, and epistemic frame theory, I argue that internship simulations can offer students opportunities to engage in the reasoning, decision-making, collaboration, and problem-solving characteristic of scientific and technical professions. Through roles such as environmental scientist, biomedical researcher, engineer, or science communicator, students encounter meaningful tasks situated in realistic workplace contexts.
Building on this pedagogical foundation, I examine how recent advances in artificial intelligence can support the design and delivery of simulated internships. First, AI can help teachers create internship experiences, reducing the time and expertise needed to develop complex simulations. Second, AI can be integrated into the simulation itself, acting as mentors, colleagues, clients, or other workplace actors with whom students interact. These developments make sophisticated internship simulations increasingly accessible to schools and universities.
The presentation concludes by outlining emerging approaches to assessing learning outcomes, including scientific reasoning, professional skills, and career awareness, while also considering the risks associated with AI-supported internship simulations. It argues that simulated internships represent a promising new direction for connecting science education with authentic professional practice and for helping students understand not only scientific knowledge but also how science is used in the world beyond the classroom.
Prof. Ming Li is currently a Distinguished Professor at the Zhejiang Key Laboratory of Intelligent Education Technology and Application, Zhejiang Normal University, China. He received his Ph.D. in Computer Science and Information Technology from La Trobe University, Australia. Following that, he completed two postdoctoral fellowships: one in the Department of Mathematics and Statistics at La Trobe University, Australia, and the other in the Department of Information Technology in Education at South China Normal University, China. He has published more than 100 papers in top-tier journals and conferences, such as IEEE TPAMI, Artificial Intelligence, IEEE TKDE, Computers & Education, BJET, NeurIPS, ICML, and AAAI. He is a member of IEEE, the China Computer Federation (CCF), the Chinese Association for Artificial Intelligence (CAAI), and the Australian Mathematical Society (AustMS). He is the recipient of the AAAI 2026 Outstanding Paper Award.
He served as the lead guest editor for the special issue "Deep Neural Networks for Graphs: Theory, Models, Algorithms and Applications" organized in IEEE TNNLS, the special session "Design and Theory of Deep Graph Learning" at the 2025 IEEE International Joint Conference on Neural Networks (IJCNN 2025), and the special session on "Recent Advances in Deep Learning for Graphs" at the 8th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science (LOD 2022). He has served as an AC/SPC/PC member for top conferences such as ICML, NeurIPS, IJCAI, KDD, and ICLR, and is also a member of the IEEE Task Force on Learning for Structured Data. He currently serves as an Associate Editor or Editorial Board Member for several international journals, including Pattern Recognition, Neural Networks, Machine Learning, Applied Intelligence, Alexandria Engineering Journal, Soft Computing, Network: Computation in Neural Systems, Neural Processing Letters, and Education and Information Technologies.
Educational data contain rich semantic information and complex relationships among learners, learning activities, and resources. Capturing these relationships is essential for understanding learning processes and providing personalized support. This talk presents recent advances in educational data analytics using graph machine learning, focusing on how graph and hypergraph representations model pairwise and higher-order interactions. Drawing on our recent research, I will discuss four applications: student performance prediction, multimodal engagement prediction, cross-modal retrieval of educational resources, and personalized course recommendation. The talk highlights the integration of large language models with signed hypergraph learning to model assessment semantics and student responses, alongside multiscale approaches that capture complementary information across modalities and learning pathways. Results on real-world educational datasets illustrate how these methods improve prediction, retrieval, and recommendation performance. By connecting advances in relational representation learning with concrete educational challenges, the talk offers insights into the design of intelligent educational systems and identifies opportunities for more reliable and personalized learning analytics.