| Invited Talk

Name of the Speaker: Dr. Sriprakash Rudrabhotla
Name of the Organizer: Prof. Krishna Jagannathan
Venue: ESB-244 (Seminar Hall)
Date/Time: 28th May 2024 (Tuesday) , 3:00 PM
Title: Online learning for service hosting

Abstract :

We consider the problem of service hosting at the edge, in which an application provider can dynamically rent resources at the edge to store its codes/libraries to serve incoming requests via short term contracts. This implies that the client can make significant cost benefits by adaptively adjusting its hosting decisions depending on the intensity of service requests, rent cost etc. We consider several online hosting policies and compare their performance for different types of arrival processes. Later, we consider the setting where the goal is to identify which amongst multiple versions of a service should be hosted, where the versions differ both in the cost they incur as well as the utility they provide to the end user. We map this problem to a variant of the best arm identification problem in the popular Multi-Armed Bandit framework, where now the different arms have correlated rewards. We derive a fundamental lower bound on the sample complexity of any policy and also propose policies with provable performance guarantees that exploit the structure of our problem

Bio: Dr. Sriprakash Rudrabhotla is a postdoc applicant (PhD from IIT Bombay 2023) who works in the area of caching and service hosting