Reinforcement Learning for Stochastic Networks, Toulouse

Session

Parallel session: Reinforcement learning and queueing I

Jun 18, 2024, 4:00 PM

A001 (ENSEEIHT)

Description

Session chair: Nicolas Gast

41. Stability and performance of multi-class queueing systems with unknown service rates: A combined scheduling-learning approach

Dr Elene Anton (Université de Pau et des Pays de l'Adour (UPPA))

6/18/24, 4:00 PM

We consider a system with $N$ different service modes handling several traffic classes, where a scheduling agent decides which service option to use at each time slot. Each service mode provides service simultaneously to all the traffic classes, at different random rates (possibly zero). Moreover, each job experiences a random slowdown when in service, which is independent among jobs and...

32. Dynamic Scheduling of a Multiclass Queue in the Halfin-Whitt Regime: A Computational Approach for High-Dimensional Problems

Ebru Kasikaralar (University of Chicago Booth School of Business)

6/18/24, 4:30 PM

We consider a multi-class queueing model of a telephone call center, in which a system manager dynamically allocates available servers to customer calls. Calls can terminate through either service completion or customer abandonment, and the manager strives to minimize the expected total of holding costs plus abandonment costs over a finite horizon. Focusing on the Halfin-Whitt heavy traffic...

46. Designing M/G/1 Scheduling Policies from Job Size Samples

Shefali Ramakrishna (Cornell University)

6/18/24, 5:00 PM

The Gittins policy is known to minimize mean response time in the M/G/1 with unknown job sizes, provided the job size distribution is known. In practice, however, one does not have direct access to the job size distribution. Motivated by this, we consider the problem of designing a scheduling policy based on a finite number of samples from the job size distribution. This is an open problem. A...