Qingsong Wen
Head of AI Research & Chief Scientist at Squirrel Ai Learning, USA
AI for Education: Leveraging Large Models for Autonomous Error Analysis and Correction
Abstract: Students frequently make mistakes while solving problems, and traditional error correction methods are both time-consuming and labor-intensive. This talk introduces an innovative Virtual AI Teacher system designed to autonomously analyze and correct student Errors (VATE). Leveraging advanced large language models (LLMs), the system uses student drafts as a primary source for error analysis, which enhances understanding of the student’s learning process. It incorporates sophisticated prompt engineering and maintains an error pool to reduce computational overhead. The AI-driven system also features a real-time dialogue component for efficient student interaction. Our approach demonstrates significant advantages over traditional and machine learning-based error correction methods, including reduced educational costs, high scalability, and superior generalizability. The system has been deployed on the Squirrel AI learning platform for elementary mathematics education, where it achieves around 80% accuracy in error analysis and shows a marked improvement in student learning efficiency.
Bio: Qingsong Wen is currently the Head of AI & Chief Scientist at Squirrel Ai Learning. His research interests include machine learning, data mining, and signal processing, especially AI for Time Series (AI4TS), AI for Education (AI4EDU), LLM & AI Agent. He has published over 120 top-ranked AI conference and journal papers, had multiple Oral/Spotlight Papers at NeurIPS, ICML, and ICLR, had multiple Most Influential Papers at IJCAI, received multiple IAAI Innovative Application Awards at AAAI, and won First Place of SP Grand Challenge at ICASSP. He also serves as Associate Editor for Neurocomputing, Associate Editor for IEEE Signal Processing Letters, Guest Editor for Applied Energy, and Guest Editor for IEEE Internet of Things Journal. In addition, he regularly serves as Area Chair of the top conferences including NeurIPS, ICML, KDD, IJCAI, ICASSP, etc.
Ping Li
Co-Founder and CEO at VecML, USA
VecML AI Data Engine for LLM Agentic Applications
Abstract: LLM agentic systems can become much more effective when agents are able to leverage the wealth of available enterprise and personal data. However, several challenges must be addressed:
- A robust and efficient database is required to handle massive and often multi-modal data, including documents, photos, videos, embedding vectors, graphs, and more.
- An efficient AI search system (often known as “RAG”) is needed to quickly and accurately retrieve relevant information from this vast data.
- Much of the data originates from edge devices such as mobile phones and personal computers, making on-device AI data technologies highly desirable.
VecML is committed to developing innovative solutions to address these challenges.
Bio: Ping Li, CEO of VecML, holds a PhD in Statistics from Stanford University. He was a recipient of the Office of Naval Research Young Investigator Program (ONR-YIP) Award when he was a faculty member at Cornell University.
Soheil Sabri
Director of Urban Digital Twin Lab – University of Central Florida, USA
Urban Digital Twin Potentials and Challenges: Enhancing Spatial Data Integration for Real-Time Analysis and Situational Awareness
Abstract: In this presentation, Dr. Sabri will discuss three major challenges of Digital Twins: fragmentation, frequency, and fidelity of spatial data for modeling, simulation, and predictive analysis. He will demonstrate how real-time analysis of spatial and temporal urban physical entities and processes—such as utility networks, urban mobility, disease transmission, and natural hazards—can be enhanced through the application of multi-scale spatial graph data models and Geospatial AI (Geo-AI). Additionally, he will discuss the value of location accuracy in inferring knowledge and insights by synchronizing Geopose standards with data streams from video recordings and mobile imagery to support personalized situational awareness and early warning systems. The presentation will also outline the potential of Multi-Agentic AI in smart and sustainable cities.
Bio: Dr. Soheil Sabri is an Assistant Professor and Director of Urban Digital Twin (UDT) Lab at the School of Modeling, Simulation, and Training (SMST) at the University of Central Florida. He holds the titles of Urban Planner and Geospatial Scientist, with a primary focus on developing Urban Digital Twins, Geospatial AI (Geo-AI), Multi-dimensional (3D/4D) Planning Support Systems, and real-time analytical tools tailored to empower planners and decision-makers with evidence-based, data-driven insights for critical infrastructure, autonomous vehicles, built environment-human interactions, and urban health. He contributes to the education, scholarship, and professional training of Digital Twins, including the Digital Twin Graduate Certification Program at the SMST, and as the lead editor of a recent Book: “Digital Twin Fundamentals and Applications”, published by Springer Nature.
He is the Ambassador and Co-chair of the Academia and Research Working Group within the Digital Twin Consortium. In addition, he has contributed substantially to the IEEE Digital Twin and Parallel Intelligence, Urban Digital Twin working group of the Open Geospatial Consortium, and the Digital Twin Task Force at the Smart Cities Council in the USA and Australia. He is an Honorary Senior Research Fellow at The University of Melbourne, Australia.
Daniel R. Isaacs
CTO & GM: Digital Twin Consortium. CSO: Object Management Group, USA
DIGITAL TWIN EVOLUTION: Catalyst for Digital Transformation
Abstract: As organizations seek competitive advantage in increasingly complex environments, digital twins provide the critical link between physical and digital realms, enabling real-time monitoring, predictive capabilities, and autonomous operations that were previously unattainable. The evolution of digital twin technology represents one of the most significant developments in the enterprise technological landscape, transitioning from Traditional to Intelligent to Generative Digital Twins.
The integration of Generative AI and multi-agent systems has fundamentally transformed digital twins from passive reflections to proactive entities capable of autonomous decision-making, scenario generation, and complex problem-solving. These intelligent digital twins now leverage multiple specialized AI agents working in concert to optimize operations, predict failures, simulate what-if scenarios, and continuous improvement from real-world feedback loops.
The Digital Twin Consortium plays a pivotal role in this ecosystem by establishing industry standards requirements, promoting interoperability, and accelerating adoption through shared best practices and reference architectures. In conclusion, compelling real-world applications where multi-agent digital twins are already delivering transformative results across manufacturing, healthcare, and other industries will be showcased. The convergence of digital twins with other emerging technologies—including artificial intelligence, edge computing, and extended reality—is accelerating this transformation, creating both opportunities and imperatives for organizational adoption and adaptation.
Bio: Dan Isaacs serves as the Chief Technology Officer and General Manager of Digital Twin Consortium. He is responsible for establishing and driving strategic technical direction and leadership to support the growth and expansion of the consortium through business development and membership initiatives. Additionally, Dan develops strategic partnerships and liaisons with other international consortiums, organizations, and alliances, to further advance the consortium’s objectives.
Dan also holds the position of as the Chief Strategy Officer for the Object Management Group (OMG). His responsibilities include developing and implementing a comprehensive strategy to unify the OMG community of consortia. Dan is responsible for driving advanced technology awareness and adoption towards accelerating sustainable digital transformation across industries, academia, government, and geographies.
Previously, as Director of Strategic Marketing and Business Development at Xilinx, Dan was responsible for emerging technologies, including AI/ML, defining, and executing the IIOT and Automotive ecosystem strategy, including responsibilities for Automotive Business Development focused on ADAS and Automated Driving systems.
Dan represented Xilinx at the Industrial Internet Consortium (IIC), leading the development of two testbeds from concept to production. Dan has over thirty years of experience working in Automotive, Industrial, Aerospace, and Consumer-based companies, including Ford, NEC, LSI Logic, and Hughes Aircraft.
An accomplished speaker, Dan has delivered keynotes and seminars and served as a panelist, and moderator at global forums and conferences Embedded World, Embedded Systems, and FPGA Conferences. He is a member of multiple international advisory boards and holds degrees in Computer Engineering from California State University and in Geophysics from ASU.
Li Deng
Chief AI Officer and Global Head of Machine Learning Vatic, USA
One simple equation that runs across neural speech recognition, computer vision & quantitative finance
Abstract: Modern AI based on deep neural nets started with speech recognition in 2009 and blossomed in computer vision in 2012 and 2015. It subsequently entered and disrupted quantitative finance in 2017. This talk briefly surveys this fascinating history, and uses the speaker’s personal research experience to review various pieces of historical neural network research that demonstrates an underlying principle which seamlessly connects successful applications of deep learning to speech recognition, computer vision & quantitative finance.
Bio: Dr. Li Deng is a Fellow of the Academy of Engineering of Canada and a Life Fellow of the IEEE. He received the 2019 IEEE SPS Industry Leader Award “For leadership in pioneering research and development on large-scale deep learning that disrupted worldwide speech recognition industry and for leadership in natural language processing and financial engineering,” and 2015 IEEE SPS Technical Achievements Award for for “Outstanding Contributions to Automatic Speech Recognition and to Deep Learning”. He co-authored a 2012 IEEE Signal Processing Magazine article with 2024 Nobel Laureate Geoff Hinton, which won the IEEE Best Paper Award.
Daoyi Dong
Australian Artificial Intelligence Institute, UTS, Sydney, Australia
AI for quantum and quantum for AI
Abstract: AI, especially machine learning, has demonstrated powerful potential to develop quantum technology. On the other hand, quantum computing has shown promising potential for developing more powerful AI. In this talk, we will briefly discuss several results on AI applications for quantum technology and quantum machine learning. In particular, we will discuss the application of reinforcement learning to quantum technology and briefly introduce quantum reinforcement learning. We will also introduce an efficient parameter initialization strategy with theoretical guarantees to enhance the trainability of parameterized quantum circuits and show that noises may make quantum kernel methods to only have poor prediction capability.
Bio: Daoyi Dong is currently a Professor and an ARC Future Fellow at the Australian Artificial Intelligence Institute, University of Technology Sydney, Australia and an Honorary Professor at the Australian National University. His research interests include quantum control and machine learning. He was awarded an ACA Temasek Young Educator Award by The Asian Control Association and is a recipient of a Future Fellowship, an International Collaboration Award and an Australian Post-Doctoral Fellowship from the Australian Research Council, and a Humboldt Research Fellowship from the Alexander von Humboldt Foundation of Germany. He is the Vice President Finance of IEEE Systems, Man and Cybernetics Society, and a member of Board of Governors, IEEE Control Systems Society. He is a Fellow of the IEEE, and a Fellow of the Australian Institute of Physics.
Tadahiko Murata
Professor at D3 Center, The University of Osaka, Japan
Human Community Digital Twin Using Synthetic Population
Abstract: In order to realize the digital twin for communities or regions in the cyber space, we need to obtain information on household members in the target area. However, it is difficult to utilize attributes of household members such as age, school, workplace, and income of household members due to privacy reason. Synthetic population method synthesizes or generates those personal data only from publicly released statistics. During the pandemic of Covid-19, those synthetic data was utilized to prepare countermeasures against Covid-19 in Japan. Since synthetic population includes household data, it can be regarded as “night-time distribution”. We now include the day-time distribution of workers such as workplace. From those information, we can develop human community digital twin that can be used for digital social experiments.
Bio: Tadahiko Murata is a Professor at D3 Center, The University of Osaka, Japan. He has classes in Division of Electronic and Information Engineering for undergraduate students and Graduate School of Information Science and Technology for graduate students. He was President of Japanese Society for Evolutionary Computation from 2020 to 2022. He is currently Vice President of Organization and Planning in IEEE SMC Society and Vice President in Japan Society for Fuzzy Theory and Intelligent Informatics. He is IEEE Fellow.
Haibin Zhu
Nipissing University, Canada
Research Innovation Based on E-CARGO in the Era of AI
Abstract: In the era of Artificial Intelligence (AI), various AI tools—particularly Large Language Models (LLMs)—are increasingly capable of performing low-level intelligent tasks such as coding and report generation. As a result, many routine jobs are at high risk of being replaced by these technologies. To meet these emerging challenges, traditional researchers and knowledge workers must equip themselves with powerful high-level modeling tools rather than relying solely on low-level programming skills.
However, AI tools also face inherent limitations, especially when it comes to innovation. For instance, knowledge-based narration remains a challenge for current LLMs, which are largely restricted to restating or reorganizing information derived from their training datasets.
E-CARGO/RBC (Environments – Classes, Agents, Roles, Groups, and Objects / Role-Based Collaboration) is a modeling methodology designed to address complex problems by enabling systematic strategy design instead of focusing on low-level implementation. RBC is a computational methodology that uses roles as the fundamental mechanism to facilitate collaborative activities. It encompasses a set of concepts, principles, models, processes, and algorithms.
The E-CARGO model and RBC methodology have evolved into powerful tools for analyzing and managing collaborative and complex systems. Research in this area has already led to significant advancements—and continues to do so—in the development, evaluation, and administration of systems across domains such as cloud computing, services, production, and organizational management.
E-CARGO/RBC has gradually matured into a foundational methodology for addressing the challenges and innovations in complex systems, including Collective Intelligence, Sensor Networks, Scheduling, Smart Cities, the Internet of Things, Cyber-Physical Systems, and Social Simulation Systems.
In this talk, the speaker will examine the need for innovative research, categorize different types of intelligence, identify the limitations of current AI tools, and introduce the RBC methodology along with its E-CARGO model. The talk will also highlight the key philosophical principles underpinning E-CARGO/RBC. The speaker welcomes questions, critical feedback, further studies, and real-world applications related to this evolving methodology.
Bio: Haibin Zhu is a Full Professor at Nipissing University, Canada. He is also an affiliate full professor of Concordia Univ. and an adjunct professor of Laurentian Univ., Canada. He has accomplished over 300+ research publications, including 60+ IEEE Transactions articles. He is a fellow of IEEE, AAIA (Asia-Pacific Artificial Intelligence Association) and I2CICC (International Institute of Cognitive Informatics and Cognitive Computing), a senior member of ACM.
He is Vice President, Systems Science and Engineering (SSE) (2023-), a co-chair (2006-) of the technical committee of Distributed Intelligent Systems, and a Distinguished Lecturer of IEEE Systems, Man and Cybernetics Society (SMCS), Associate Editor (AE) of IEEE Transactions on SMC: Systems (2018-), IEEE Transactions on Computational Social Systems (2018-), IEEE Systems Journal (2024-), Frontiers of Computer Science (2021-), IEEE Canada Review (2017-), and Deputy Editor-in-Chief of Artificial Intelligence Science and Engineering (2025-). He was General Chair: E-CARGO 2025, China, 2024, China, 2023, online, ScalCom 2023, UK; ISEEIE 2023, Singapore, SPCS 2022, China, ICCSIT 2021, France, and Program Chair: ICFTIC 2025, 2024, China, CSCWD 2020, CSCWD 2018, China, ICNSC 2019, and CSCWD 2013, Canada.
He is the founding researcher of Role-Based Collaboration and the creator of the E-CARGO model. He has offered 38 keynote speeches for international conferences and 93 invited talks internationally.