[Computational Learning Theory Team Seminar] Talks on Machine Learning for Mobility and Sensing

Two invited research talks on machine learning for mobility and sensing, held at Kyushu University's Ito Campus in Fukuoka and streamed on Zoom for registered participants. The session runs from 13:30 to 15:00, with roughly 45 minutes per talk including Q&A. Prof. Rizk Hamada (Osaka University) presents "AI-Driven Spatial Intelligence for Cyber-Physical Systems", covering how sensing data is turned into spatial representations, predictions, and context-aware actions. The talk focuses on the underlying learning problems: generalizing from limited labeled data, staying robust to device heterogeneity and environmental change, and learning representations from large unstructured spatial data, alongside privacy-preserving human sensing and explainable reasoning for real-time robotic decisions. Prof. Ahmed Gomaa (EJUST, Egypt) follows with "Deep Learning Revolution in Remote Sensing: From Pixels to Life-Saving Disaster Response", on why state-of-the-art models break down in unseen remote-sensing environments. He addresses the generalization gap through domain adaptation, multimodal sensor fusion, and edge optimization, arguing that deployable systems need to be trustworthy and generalizable, not only accurate.