Tomoaki Otsuki

Tomoaki Otsuki (Ohtsuki) holds B.E., M.E., and Ph.D. degrees in Electrical Engineering from Keio University, Japan. He is a professor at Keio and an Adjunct Professor at the University of Houston. He previously worked at UC Berkeley, and his research covers wireless and optical communications, signal processing, and information theory. Dr. Ohtsuki has received numerous awards, including the Inoue Research Award, IEEE awards, and Best Paper Awards at various conferences. He has published over 320 journal and 540 conference papers.

He has served in leadership roles in IEEE and other societies, including as Director of IEEE ComSoc Asia Pacific, and as editor for several flagship journals. He has also been a keynote speaker and session chair at many international conferences. He is a Fellow of IEICE, AAIA, and a Senior Member of IEEE, and a member of Japan’s Engineering Academy.

Keynote: Wi-Fi Sensing for Human Activity Recognition: Leveraging Deep Learning and Data Mining

This keynote presentation explores the emerging domain of human activity recognition (HAR) utilising Wi-Fi sensing technologies. It specifically examines the application of channel state information (CSI) combined with deep learning methodologies to detect and classify human behaviours. Initially, the presentation provides an overview of Wi-Fi sensing, emphasising its potential as a non-invasive, cost-effective, and accurate approach to human activity recognition. Subsequently, it discusses the challenges associated with implementing Wi-Fi-based HAR, including issues related to subject variability, environmental fluctuations, and data quality concerns. The presentation then delineates proposed methodologies for enhancing the robustness of activity recognition systems through advanced deep learning techniques. These include sophisticated feature extraction from Wi-Fi signals and the employment of cutting-edge models aimed at improving classification accuracy. The discussion extends to addressing models’ generalisation capabilities across different subjects and environmental scenarios, illustrating their performance with diverse user profiles and settings. A central focus is placed on developing models that are resilient to real-world conditions characterised by noise and signal degradation. Furthermore, the presentation considers the development of resource-efficient, lightweight models suitable for deployment on constrained devices and explores multi-label classification strategies to identify multiple activities concurrently. Overall, this work aims to advance the field of Wi-Fi-based HAR by addressing key technical challenges and proposing solutions for practical, real-world applications.

Nitin H Vaidya

Nitin Vaidya is the Robert L. McDevitt, K.S.G., K.C.H.S. and Catherine H. McDevitt L.C.H.S. Chair Professor of Computer Science at Georgetown University, where he served as the Department Chair during 2018-24. His current research interests are in the area of distributed algorithms, and previously, he has worked on wireless networks. He received a Ph.D. from the University of Massachusetts at Amherst. He previously served as a Professor and Associate Head in Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign. He has co-authored papers that received awards at several conferences, including SSS, ACM MobiHoc and ACM MobiCom. He is a fellow of the IEEE. He has served as the Chair of the Steering Committee for the ACM PODC conference, as the Editor-in-Chief for the IEEE Transactions on Mobile Computing, and as the Editor-in-Chief for ACM SIGMOBILE publication MC2R.

Keynote : Fault-Tolerant Distributed Optimization and Learning

Consider a network of agents wherein each agent has a private cost function. In the context of distributed machine learning, the private cost function of an agent may represent the “loss function” corresponding to the agent’s local data. The objective here is to identify parameters that minimize the total cost over all the agents. In machine learning for classification, the cost function is designed such that minimizing the cost function should result in model parameters that achieve higher accuracy of classification. Similar problems arise in the context of other applications as well.

Our work addresses privacy and security (or fault-tolerance) of distributed optimization with applications to machine learning. In privacy-preserving machine learning, the goal is to optimize the model parameters correctly while preserving the privacy of each agent’s local data. In fault-tolerance, the goal is to identify the model parameters correctly while tolerating adversarial agents that may be supplying incorrect information. When a large number of agents participate in distributed optimization, security compromise or failure of some of the agents becomes increasingly likely. This talk will discuss some fault-tolerant algorithms for distributed optimization, with applications to learning.

Grace Hui Yang

Dr Grace Hui Yang is an Associate Professor in the Department of Computer Science at Georgetown University, where she leads the InfoSense (Information Retrieval and Sense-Making) group. She earned her PhD from Carnegie Mellon University in 2011. Her research interests include deep reinforcement learning, conversational AI, dynamic information retrieval, and privacy-preserving information retrieval. Previously, she focused on question answering, ontology construction, and multimedia information retrieval. Dr Yang has received support from the Defence Advanced Research Projects Agency and the National Science Foundation, including the NSF CAREER Award. She has co-chaired several workshops and conferences, including SIGIR Doctoral Consortia, and has served on the editorial board of the Information Retrieval Journal. Additionally, she co-organised the TREC Dynamic Domain Track from 2015 to 2017 and led SIGIR workshops on privacy-preserving information retrieval from 2014 to 2016. For more information, visit her homepage: http://infosense.cs.georgetown.edu/grace/.

Keynote: Reinforcement Learning for Agentic Information Access: From Proactive Communication to Trustworthy Intelligence

As artificial intelligence systems evolve from passive responders to proactive communicators, the frontier of information access is shifting toward agentic intelligence, where AI agents can autonomously seek, verify, and communicate knowledge. Reinforcement Learning (RL) offers a powerful foundation for training such systems, enabling them to balance exploration, communication, and trust. This talk examines how RL, especially offline, hierarchical, and goal-conditioned formulations, can empower conversational and information-access agents to act inquisitively and responsibly within complex environments. I will present our recent work on Reward-on-the-Line offline RL for domain-specific dialogue agents and Inquisitive Conversational Agents that learn to ask meaningful questions, uncover missing information, and reason with evidence. Together, these efforts point toward a future where reinforcement learning supports not only effective communication but also trustworthy information ecosystems, key foundations for ambient intelligence and responsible AI.