Invited Talk at Muroran Institute of Technology (Co-organized)
An invited talk by Associate Professor Kazuya Sakai, Tokyo Metropolitan University, Japan, will be held on July 8, 2026, in Room R205, Education & Research Building No. 8, Muroran Institute of Technology, 27-1 Mizumoto-cho, Muroran, Hokkaido, 0508585 Japan. Associate Professor Kazuya Sakai will share some interesting ideas about Distributed Multi Armed Bandits for Mobile Social Networks.
CO-ORGANIZED BY:
IEEE Muroran Institute of Technology Student Branch (SB)
IEEE Systems, Man, and Cybernetics Society Muroran Institute of Technology Student Branch Chapter
IEEE Computer Society Muroran Institute of Technology Student Branch Chapter
IEEE Sapporo Section Young Professionals (YP)
The Center for Computer Science (CCS), Muroran Institute of Technology
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
- 27-1 Mizumoto-cho, Muroran, Hokkaido
- Muroran, Hokkaido
- Japan 0508585
- Building: R205
- Starts 08 July 2026 01:10 AM UTC
- Ends 08 July 2026 01:25 AM UTC
- No Admission Charge
Speakers
Associate Professor Kazuya Sakai of Tokyo Metropolitan University, Japan
Distributed Multi Armed Bandits for Mobile Social Networks
In this talk, I will talk about federated multi-armed bandits (MABs) for mobile social networks. The first half of this talk is dedicated for a brief introduction to MABs, which is widely used reinforcement learning methodology for decision-making processes. In MAB, an agent's goal is to maximize the long-time reward by balancing exploiting the current knowledge and exploring actions. Then, in the second half of this talk, I will present distributed and federated MAB for mobile social networks, where a set of agents learn action values by locally observing rewards and by exchanging estimated action values with others. The key component of the proposed federated bandit algorithm includes the agent's centrality metric, called weighted-connectivity, which quantifies the influence of individual agents to other agents. Then, I design the weighted-connectivity-based upper confidence bound (WC-UCB) algorithm that exploits limited federated opportunities by prioritizing the action value estimates of the agents with a high centrality score. The performance is evaluated by simulation using real mobility traces, such as CRAWDAD and CDC, and WC-UCB outperforms the state-of-the-art bandit algorithms.
Biography:
Kazuya Sakai received his Ph.D. degree in Computer Science and Engineering from The Ohio State University in 2013. He is currently an associate professor at the Department of Computer Science, Tokyo Metropolitan University. His research interests are in the area of information and network security, wireless and mobile computing, and distributed algorithms. He received the IEEE Computer Society Japan Chapter Young Author Award 2016. He is a member of the IEEE and ACM.
Address:Japan