When: Thursday, May 4, 2023 08:45 PM-10:00 PM Indiana (East) | Friday, May 5, 2023 07:45 AM-9:00AM Vietnam time
The talk will start after 15 minutes of “coffee break”
Register in advance for this meeting:
https://iu.zoom.us/meeting/
After registering, you will receive a confirmation email containing information about joining the meeting.
Speaker: Prof. Qin Zhang (Indiana University Bloomington)
Qin Zhang is an associate professor in the Department of Computer Science at Indiana University Bloomington. He received his Ph.D. from the Hong Kong University of Science and Technology. Prior to IUB, he spent a couple of years as a postdoc at Theory Group, IBM Almaden Research Center, and the Center for Massive Data Algorithmics, Aarhus University. He is interested in algorithms for big data (in particular, communication-efficient distributed computation and streaming/sketching algorithms) and the theoretical foundations of machine learning (in particular, collaborative learning and distributed learning). He has published extensively in premier conferences in theory/algorithms, databases, machine learning, and data mining. He is the recipient of several NSF grants (including a CAREER Award) and the Best Paper Award at SPAA 2017.
Title of the Talk: Collaborative Learning with Limited Communication
Abstract: In this talk, I will introduce my recent work on collaborative (reinforcement) learning (CL), in which multiple agents work together to learn an objective function. We are particularly interested in a scenario in which agent communication is very expensive. Our goal is to identify the tradeoffs between the speedup of the collaboration and the communication cost among the agents. We convey the following messages using a basic problem in bandit theory as a vehicle: (1) adaptive CL is more powerful than non-adaptive CL; (2) CL with non-IID data is harder than that with IID data; and (3) problems incomparable in the single-agent learning model can be separated in the CL model.