|
Title: Bayesian inference and inverse tasks with queueing networks. Speaker: Dr. Iker Perez, University of Nottingham, UK Date and Time: Wednesday 29 August, at 12:00 noon Venue: IEOR Teaching Room (ground floor) |
 |
Abstract: Queueing networks are systems of theoretical interest that find widespread use in performance evaluation tasks with interconnected resources. These networks are of great practical value in domains as diverse as operations research, computing and telecommunications. However, the underlying mathematical constructions often given rise to complex stochastic models that pose impediments to foundational statistical studies. As a consequence, there exist few relevant approaches for transient inference and uncertainty quantification tasks for queueing systems.
In this talk, I will first provide a basic overview of the inverse inferential challenge. By means of common accessible examples, I will discuss its relation with analogue statistical tasks in related mathematical domains. Next, I will present a network model augmentation, and summarise key results that enable the construction of an approximate variational Bayesian framework, suitable for inference with general-form queueing systems. This will be followed by a few examples discussing uncertainty quantification tasks for network service rates.
Speaker profile: Dr Iker Perez is Assistant Professor in Statistics at the Horizon Digital Economy Research Institute at the University of Nottingham, UK.