Asst. Prof. Sinan CHEN
Kobe University, Japan

 

Biography: Sinan Chen, Ph.D., received his M.S. degree in System Informatics and Ph.D. degree in Computational Science from Kobe University, Japan, in 2019 and 2021, respectively. He was a JSPS Postdoctoral Research Fellow at Kobe University in 2022 and is currently an Assistant Professor at the Center for Mathematical and Data Sciences, Kobe University.
His research interests include human-centered AI, multimodal behavioral sensing, computer vision, temporal behavior modeling, edge AI, digital biomarkers, and context-aware intelligent systems. His recent work focuses on real-world human behavior understanding using facial, hand, body, gait, and everyday activity data, with applications in digital health, smart environments, and personalized intelligent systems. His broader research goal is to develop multimodal behavioral intelligence for reliable and adaptive human-centered AI.
He is a member of the IEEE, ACM, Sigma Xi, CAAI, CCF, IEICE, and IEEJ.

 

Title：Multimodal Behavioral Intelligence: Context-Aware Vision, Temporal Modeling, and Edge AI for Real-World Human-Centered Systems

Abstract
Recent advances in computer vision, multimodal sensing, and edge artificial intelligence have greatly improved the ability of intelligent systems to recognize human faces, gestures, poses, activities, and physiological or behavioral signals. However, reliable real-world human understanding requires more than recognizing isolated observations. Human behavior is inherently temporal, context-dependent, multimodal, and highly variable across individuals and environments.
In this invited talk, I will present a research perspective on multimodal behavioral intelligence, aiming to move from isolated recognition toward context-aware and adaptive understanding of human behavior over time. Representative examples from facial dynamics, hand and finger movements, body motion, gait, and everyday activity sensing will illustrate how heterogeneous visual and behavioral signals can be integrated to characterize meaningful behavioral patterns and individual changes.
Particular attention will be given to three challenges: temporal modeling of behavioral dynamics, contextual integration across sensing modalities, and reliable deployment through lightweight and edge-based AI. I will also discuss how privacy-aware processing, uncertainty, communication constraints, and individual variability affect the transition from laboratory models to real-world intelligent systems.
Finally, I will outline a broader research direction in which multimodal sensing, signal representation, temporal modeling, and adaptive AI are jointly designed to support human-centered systems across digital health, smart environments, human–computer interaction, and personalized intelligent services.

 

 

 

 

 

•