We are pleased to invite you to attend the Department of Management Science and Engineering seminar series in the Faculty of Engineering at the University of Waterloo!
Our seminar series often accommodates a variety of talks such as research-oriented, Ph.D. student, and practitioner-oriented talks. Topics in the series include, among other, energy systems, optimisation, logistics and supply chain, healthcare, information and communications technologies, and socio-technical systems.
The audience, usually 15-25 in size, is primarily composed of faculty and graduate students from the Department of Management Science and Engineering, although other faculty, students and industrial professionals frequently attend.
The department seminar series' coordinator is Kejia Zhu.
Degree Requirements
- MASC students are required to attend 8 seminars for degree completion
- MMSc students are required to attend 4 seminars for degree completion
- Students must attend the entire seminar (at least 45 minutes) to receive credit
Note: To receive credit, students can only attend MSE Department Research Seminars.
PhD Seminars | MASc Seminars | Past Seminars
Department Research Seminars:
Uncertainty management in radiotherapy via AI & optimization
Speaker: Dr. Akra Roy
Alvarez College of Business
Date: July 19th 2:00pm-3:00pm EDT
E7 Faculty Hall (E7-7303, E7-7363)
Abstract: In radiotherapy, uncertainties can reduce the quality of treatments; deterministic treatments can lead to suboptimal outcomes for the patient. Ultimately, this can cause over- or under-treatment, leading to dysfunctional organs from excessive radiation exposure or tumour metastasis from poor coverage and a failed treatment.
In this talk, we will explore methods from the areas of optimization under uncertainty and artificial intelligence (AI) that can account for two sources of uncertainties, some examples include:
-
Radio resistance
-
Contouring errors.
While standard AI methods and (naïve-) robust treatment planning can overcome this, they can also produce overly conservative or suboptimal decisions, especially when there are unpredictable changes during the treatment.
To overcome this, a predictive-prescriptive adaptive framework is proposed that can adapt to updated information during the treatment. Using this framework, we will explore the value of multiple diagnostics to update the treatment and derive optimal intervention policies that account for time-dependent uncertainties.
Uncertainties in the airspace: from distributionally robust air traffic flow management to stochastic routing problems for drone package delivery systems
Speaker: Professor Max Z. Li
University of Michigan
Date: February 21st 2:00pm-3:00pm EDT
E7 Faculty Hall (E7-7303, E7-7363)
Abstract: The US National Airspace System (NAS) is becoming increasingly interconnected, and data driven. The Information-Centric NAS vision laid out by the US Federal Aviation Administration encapsulates this, with focus on growing operations, infrastructure for traffic management, and safety assurance, all to increase predictability and resilience. However, given the myriad sources of dynamic uncertainty inherent within the NAS, predictions must incorporate not just this uncertainty, but the potential that the uncertainty itself is incorrectly modeled or characterized. To this end, I will discuss applications of distributionally robust optimization to a family of integer optimization problems for air traffic management decision-making (which flights to delay, and for how long). I will then move to discuss the impact of airspace and environmental uncertainties (e.g., stochastic winds) on new airspace entrants such as drones and other un-crewed aerial systems (UAS) and explore methods for characterizing and accounting for uncertainties.