Math Modelling - INVITED LECTURES
Winter 2021
Mathematical models in Finance
Letitia Golubitsky - March 23, 1pm-2.20 pm Lecture Link
Forecasting interest rate accurately is one of the hardest problem in quantitative mathematics in financial markets. There are numerous existing mathematical techniques which are currently employed by banks to forecast interest rates. In this talk I will attempt to present several competing mathematical approaches to model interest rate . Monte Carlo techniques are heavily used in our simulated exposure to changes in interest rates .
Model calibration considered a state-of-art component of the mathematical modeling in finance relies on advanced statistical and optimization methods.
BIO: Letitia Golubitsky has an M. Sc. in Mathematics from Queen's University, a Ph. D. in Mathematics from University of Western Ontario and a Master in Financial Mathematics from McMaster. She worked in the financial world for the last 9 years and has extensive expertise in Counterparty credit risk modelling in Model Validation and Model Development across 3 big Canadian Banks. She has working experience on Internal Model Method (IMM) application for CCR and FRTB for Market Risk. Expert matter in Commodity modelling and Commodity pricing derivatives working with multiple stakeholders: traders, developers, regulators and audit.
Vaccine to Market: Development and Research
Thomas Shin - March 25, 2021, 1pm- 2.20 pm Lecture link
Vaccine investment and innovation have become a prominent and frequent topic within the public, media, medical, and government discourse since the emergence of COVID-19. What may be under-reported is how vaccines are developed, studied, processed, distributed, and researched to ultimately reach the public (e.g. physician’s office, pharmacy, etc.). In this talk, I will provide a brief overview of the vaccine life cycle and the different types of vaccine-related research conducted from an industry perspective.
BIO: Thomas Shin is an academically trained epidemiologist and statistician currently leading the epidemiology and real-world evidence (RWE) generation efforts for the medical team at Sanofi Pasteur Canada. Before joining Sanofi Pasteur, Thomas worked as a senior health economist at Cancer Care Ontario (CCO), where he advised decision-makers in the New Drug Funding Program (NDFP) and the Pharmacoeconomics Research Unit (PRU). He was later recruited by the Ministry of Health and Long-Term Care (MOHLTC), where he worked as a senior economist for the Ontario Provincial Drug Program (now called Drugs and Devices) and the Division of Policy and Strategy. Thomas received his HBSc, MA, and MPH in epidemiology at the University of Toronto and will be finishing his MA-GDFE (2021) in applied statistics and financial engineering through the Department of Math and Statistics and the Schulich School of Business at York University.
Winter 2020
Heterogeneous Graph Clustering Using a Pointwise Mutual Information Criterion
Oleg Golubitsky
Graph mining is a rapidly growing research area, owing its success to the fact that graphs frequently offer a more flexible and intuitive representation of real-world phenomena than vector spaces. Most of the current graph clustering algorithms are designed for homogeneous graphs, in which nodes have the same type. However, in many practical settings, information about node types is readily available and too valuable to be discarded up front.
We will discuss a graph clustering method that uses connections to shared neighbours of different types directly, in order to assign higher scores to sets of nodes that form "good" clusters, i.e., are statistically more likely to result from a hidden common cause, rather than having the observed common neighbours by accident. The method is local, i.e., when the graph changes slightly, a small amount of computation is needed to update the clusters and their pointwise mutual information scores.
(joint work with Pushkarini Agharkar and Dake He)
From planets to population health
Edward Thommes
Modeling of complex systems can take many forms, and a career in this field may start in one place yet end up somewhere very different. I will discuss computationally-focused research I have conducted in two very different areas, namely planetary astrophysics and biomathematics. The former centers on the processes by which a young protostar's disk transforms itself into a planetary system, how these processes lead to the huge variety of discovered exoplanets, and how our own Solar System fits into the picture. The latter involves inferring the burden and past history of infectious diseases from surveillance and health system data, and using this to make predictions about the future, including the effect that future interventions (e.g. new vaccines) might have. Although on the face of it these two areas have little in common, some computational methods and approaches can be applied to both. Others, however, absolutely cannot. I will provde examples of either case.
Vaccine research using large linkable databases
Jeff Kwong
Health administrative data have been used for decades for pharmacoepidemiologic research, but their use for research on vaccine-preventable diseases has generally been more limited. In this talk, I will provide some examples of how we have used linked databases (e.g., laboratory data, health administrative data) in Ontario to conduct epidemiologic studies on vaccines and vaccine-preventable diseases.
Oleg Golubitsky graduated from Moscow State University and University of New Brunswick with PhDs in Math and Computer Science, worked as an assistant professor at UNB and as a postdoc at the University of Pisa, Queen's and Western University in the areas of differential algebra, symbolic computation and pattern recognition. He then joined Google as a software engineer, first to develop products that optimize performance of online ad campaigns, then to ensure that they comply with ad policies and are safe for users. Oleg enjoys projects that involve coding, data analysis, research and experimentation and is going to talk about one of them.
Edward Thommes is Director of Vaccine Epidemiology and Modeling at Sanofi Pasteur, adjunct professor in the Department of Mathematics and Statistics at University of Guelph as well as York University, and an Affiliate Member of the Waterloo Institute for Complexity and Innovation. He holds a BSc in Physics from University of Alberta (1994), and a PhD in Astrophysics from Queen’s University (2000). His graduate and postdoctoral research centered on the formation and dynamics of the Solar System and exoplanetary systems. His current research interests include biomathematics, epidemiology, the conduct of real-world evidence studies, advanced analytics and health economics.
Jeff Kwong is an epidemiologist, a specialist in public health and preventive medicine, and a family physician. He is the Program Leader of the Populations and Public Health Program at ICES (a research institute that houses a large array of linkable health-related databases), a Scientist at Public Health Ontario, and a Professor at the University of Toronto. As a Clinician-Scientist, he practises family medicine one day per week and devotes the rest of his time to research and teaching at the interface between primary care and public health. His research interests include infectious diseases epidemiologic research using large linkable databases, influenza vaccine and vaccination program evaluation, and assessing the health and economic burden of infectious diseases.