Thomas Ploetz

Georgia Institute of Technology, United States

Title: “We’ve come a long way: Human Activity Recognition in the Age of (Modern) AI”

Abstract: Human activity recognition (HAR) utilizing motion sensing platforms that are either body-worn or integrated into the environment has been a fundamental pillar of ubiquitous and pervasive computing since the establishment of this research field. HAR has made significant progress. The history of HAR research can be effectively categorized into three epochs: (1) The early years were characterized by explorations on how HAR can effectively be integrated into various Ubicomp application domains, establishing methods for sensing and standardized forms of machine learning-based sensor data analysis. (2) With the advent of end-to-end learning, HAR became more mature, and the focus shifted from exploration to “professionalization,” i.e., to seriously pushing towards robust and accurate recognition, particularly for challenging deployment scenarios prevalent in many Ubicomp applications. (3) With the emergence of techniques such as generative methods, foundation models, and the transition to true multi-modality, we have entered the third era: HAR in the age of modern AI. This era promises not only solving the notoriously challenging problem of sensor-based human activity recognition but also facilitating the exploration of new frontiers that extend beyond the mere distinction of human activities. In this keynote, I will contextualize current trends in sensor-based HAR, specifically modeling and analysis methods, within new frontiers, new challenges, and new directions in both research and practical deployments of HAR. These frontiers include the shift to true multimodality, continual learning, and representation-based abstraction, which facilitates generalized HAR deployments. The widespread adoption of new, AI-based modeling comes with challenges such as the requirement for proper representations of sensor data streams, the need to revise established evaluation protocols due to potential memorization, and the surprising difficulty of utilizing language supervision “out of the box.” I will discuss approaches to tackle such challenges. I will conclude the keynote with a discussion on the future prospects: a promising era where the community can focus on more intricate problems with substantial potential for HAR deployments in real-world application scenarios that transcend fitness tracking and sleep analysis.

Speaker Details: Thomas Ploetz is a Computer Scientist with expertise and decades of experience in Pattern Recognition and Machine Learning research (PhD from Bielefeld University, Germany). He works as a Professor of Computing at the School of Interactive Computing at the Georgia Institute of Technology in Atlanta, USA, where he leads the Computational Behavior Analysis research lab (cba.gatech.edu). There he is also the Associate Chair for Graduate Studies. His research agenda focuses on applied machine learning, that is developing systems and innovative sensor data analysis methods for real world applications. Primary application domain for his work is computational behavior analysis where he develops methods for automated and objective behavior assessments in naturalistic environments, thereby making opportunistic use of ubiquitous and wearable sensing methods. Main driving functions for his work are "in the wild" deployments and as such the development of systems and methods that have a real impact on people's lives. Thomas is a passionate educator teaching (very) large classes on Artificial Intelligence and Mobile, and Ubiquitous Computing on a regular basis at Georgia Tech, and worldwide through guest lectures and keynotes. Thomas has been very active in the mobile and ubiquitous, including wearable computing community. He is co-editor in chief of the Proc. of the ACM on Interactive, Mobile, Wearable, and Ubiquitous computing technology (IMWUT), has twice been co-chair of the technical program committee of the International Symposium on Wearable Computing (ISWC), and was general co-chair of the 2022 Int. Joint Conf. On Pervasive and Ubiquitous Computing (Ubicomp). Thomas is a Distinguished Member of the ACM.

Chiara Boldrini

Institute for Informatics and Telematics, CNR, Italy

Title: Intelligence Without a Center: The Promise of Fully Decentralised AI

Abstract: Fully decentralised learning marks a shift in how we think about intelligence itself. By keeping data where it is created and enabling knowledge to flow through networks of peers, it promises AI that is more private, sovereign, and responsive to the world around it. Yet with this promise comes a new set of questions: How does intelligence emerge from the fabric of network topology? How can systems remain resilient when data or communication falter? And what happens when knowledge must be distilled from imperfect or low-quality inputs? In this keynote, we explore how decentralisation transforms these challenges into opportunities, pointing to a future where AI is not a single centralised engine, but a living, adaptive ecosystem.

Speaker Details: Chiara Boldrini is a Senior Researcher at IIT-CNR and head of the AI & Data Science lab of the Ubiquitous Internet research unit. Her research interests are in human-centric decentralized AI, causal learning in pervasive systems, human behavioral/cognitive models for the analysis and design of online social networks/Metaverse. She is the IIT-CNR co-PI for the National Extended Partnership in Artificial Intelligence (FAIR) and for the PNRR ICSC project, and she was involved in several EC projects since FP7. She currently holds the position of Editor-in-Chief for Special Issues at Elsevier Computer Communications. She is serving as General Chair of IEEE PerCom'26 (A* CORE Ranking) and, over the years, has been on the organizing committee of several IEEE and ACM conferences/workshops, including IEEE PerCom and ACM MobiHoc. Recently, she has served in the TPC of AAAI ICWSM (as senior PC member), The Web Conference, ECML-PKDD, WSDM, and MobiHoc, among several others.

Stephan-Sigg

Aalto University, Finland

Title: Coming Soon

Abstract: Coming Soon

Speaker Details: Stephan Sigg is a Professor at Aalto University in the Department of Information and Communications Engineering. His research interests include the design, analysis and optimisation of algorithms for distributed and ubiquitous systems. Especially, his work covers proactive computing, distributed adaptive beamforming, context-based secure key generation and device-free passive activity recognition. Stephan is an editorial board member of the Elsevier Journal on Computer Communications and has been a guest editor for the Springer Personal and Ubiquitous Computing Systems Journal. He has served on the organizing and technical committees numerous prestigious conferences including IEEE PerCom, ACM Ubicomp.