Talk on "Toward A Pessimistic-Optimistic Framework for Constrained Bandit Learning" by Dr. Ning Lu from Queen's University
We are very pleased to announce the following talk by Dr. Ning Lu from Queen's University, who visited NII as an intern student in 2009.
You are very welcome to join us.
Toward A Pessimistic-Optimistic Framework for Constrained Bandit Learning
The online optimization of time-varying quantities with unknown statistics is a common issue in modern engineering problems, particularly in communication networks. The multi-armed bandit (MAB) problem is a suitable model for many of these problems, where the player must choose actions that yield random rewards from an unknown distribution.
In many cases, these problems also have arm selection constraints, where playing or not playing an arm expends some budgeted resource. Recent work has introduced pessimistic-optimistic algorithms to handle these constraints and track constraint violations using virtual queues. However, many engineering problems require additional modeling considerations, such as delayed feedback or arm switching costs, that may not fit neatly within this framework.
In this talk, we will introduce an extensible framework for pessimistic-optimistic algorithms that can accommodate these additional considerations. We will also introduce our recent work exploring constrained bandit learning with delayed feedback and with switching costs, along with some novel results that more broadly impact constrained bandit learning problems.
Dr. Ning Lu, Assistant Professor in the Department of Electrical & Computer Engineering at Queen's University
Dr. Ning Lu is an Assistant Professor in the Department of Electrical & Computer Engineering at Queen's University. He is also a Tier 2 Canada Research Chair in Future Communication Networks.
Dr. Lu received the B.Eng. (2007) and M.Eng. (2010) degrees from Tongji University, Shanghai, China, and Ph.D. degree (2015) from the University of Waterloo, Waterloo, ON, Canada, all in electrical engineering.
Prior to joining Queen's University, he was an assistant professor in the Department of Computing Science at Thompson Rivers University, Kamloops, BC, Canada. From 2015 to 2016, he was a postdoctoral fellow with the Coordinated Science Laboratory, University of Illinois at Urbana-Champaign. He also spent the summer of 2009 as an intern in the National Institute of Informatics, Tokyo, Japan.
His current research interests include real-time scheduling, distributed algorithms, and reinforcement learning for wireless communication networks. He has published more than 40 papers in top IEEE journals and conferences, including IEEE/ACM Transactions on Networking, IEEE Journal on Selected Areas in Communications, ACM MobiHoc, and IEEE INFOCOM, etc.
He was a recipient of John R. Evans Leaders Fund. He received a best paper award at the 2014 IEEE GLOBECOM. He has served as journal guest editor, TPC member of major IEEE conferences, and reviewer of refereed journals.
14:00 - / Tuesday, July 11th, 2023
Hybrid(Onsite and Online)
If you would like to join, please contact by email.
Email : kei [at] nii.ac.jp
Yusheng Ji Laboratory