Pengyang Wang

University of Macau, pywang@um.edu.mo

Speaking Topics

• Towards Sustainable, Resilient, and Participatory Urban Planning: An AI-Driven Approach

Speaker's Notes

Our urban environments stand at a critical crossroads, facing unprecedented challenges from climate change and resource scarcity to social inequity. As traditional planning methodologies buckle under the weight of these complex issues, this keynote illuminates a transformative path forward, leveraging Artificial Intelligence as the catalyst for creating more sustainable, resilient, and participatory cities. Drawing from pioneering recent studies, we will journey from theory to practice, showcasing tangible AI applications that tackle critical issues—from designing energy-efficient buildings and preventing sewer overflows to revolutionizing participatory planning with human-in-the-loop feedback. This talk will offer a clear-eyed vision of both the immense opportunities and the profound challenges that lie ahead in this AI-driven urban frontier.

Guang Wang

Florida State University, guang.wang@fsu.edu

Speaking Topics

• Efficient Generative AI for Synthetic Human Activity Trace Generation

Speaker's Notes

Fine-grained human activity traces (HATs) are crucial for a wide range of real-world applications, including human mobility modeling, next point-of-interest (POI) recommendation, business location optimization, trip purpose inference, and pandemic intervention. However, increasing concerns over data privacy have significantly restricted access to authentic large-scale HATs. Fortunately, r ecent advances in generative AI open new opportunities to synthesize realistic yet privacy-preserving HATs that can support diverse applications. Despite this promise, two key challenges remain. (i) HATs (e.g., user-level POI check-in traces) are highly irregular and dynamic. (ii) Most generative models are computationally intensive and resource-demanding. To address these challenges, we propose SynHAT, a coarse-to-fine HAT synthesis framework based on a novel spatio-temporal denoising diffusion model. Extensive experiments on four real-world datasets show the superior performance our SynHAT in terms of different spatial and temporal metrics.

Zhe Jiang

University of Florida, zhe.jiang@ufl.edu

Speaking Topics

• New Frontiers of Generative AI for Geospatial Data: Applications in Hydrology and Oceanography

Speaker's Notes

Over the past decade, generative AI techniques have revolutionized computer vision and natural language processing. There is growing anticipation for similar breakthroughs in scientific domains, driven by pressing societal challenges such as national water resource management, energy and food security, and climate change mitigation and adaptation. However, geospatial data present unique challenges for existing AI models, including spatiotemporal autocorrelation and heterogeneity, long-range and multi-scale dependencies, the existence of physical knowledge and constraints, and the paucity of ground truth. In this talk, I will present my recent research aimed at addressing these challenges, including (1) geospatial AI for observation-based flood inundation mapping in hydrology, and (2) AI surrogates to accelerate coastal ocean circulation modeling in physical oceanography.

Ranga Raju Vatsavai

North Carolina State University, rrvatsav@ncsu.edu

Speaking Topics

• GeoAI in the Era of Foundation Models: Challenges and Opportunities

Speaker's Notes

Decades of research have culminated in significant advances in artificial intelligence (AI), particularly deep learning (DL) and foundation models (FM). These technologies hold immense potential for addressing critical global challenges, such as climate change mitigation, food security, smart city design, and resource optimization. A common thread uniting these issues is their inherent spatial and temporal nature. Remote sensing data is a prime example of spatial big data, yet a major challenge persists: with over 50% of the Earth's surface covered by clouds at any given time, cloud contamination severely degrades the performance of downstream tasks like image segmentation, scene understanding, and land cover classification. Furthermore, the integration of multimodal data presents additional challenges related to differing spatial and spectral resolutions. This talk will explore the evolution of geospatial artificial intelligence solutions for these problems, beginning with techniques for reconstructing cloud-contaminated regions and progressing to recent breakthroughs in geospatial foundation models and their diverse applications. The presentation will conclude with a discussion of pressing open challenges in the field.