SCIENTIFIC ACTIVITIES - Fields Institute - University of Toronto
SCIENTIFIC ACTIVITIES - Fields Institute - University of Toronto
SCIENTIFIC ACTIVITIES - Fields Institute - University of Toronto
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Distinguished Lecture Series in Statistical Sciences<br />
JIANQING FAN IS WELL-KNOWN FOR HIS WORK<br />
in financial econometrics, computational biology, semiparametric<br />
and nonparametric modeling, and other aspects <strong>of</strong> statistical<br />
theory and methodologies. He is the Frederick I. Moore Class<br />
<strong>of</strong> 1918 Pr<strong>of</strong>essor in Finance, Director <strong>of</strong> the Committee <strong>of</strong><br />
Statistical Studies at Princeton, the Past President <strong>of</strong> the <strong>Institute</strong><br />
<strong>of</strong> Mathematical Statistics, and winner <strong>of</strong> the 2000 COPSS<br />
Presidents’ Award.<br />
In early May,<br />
Jianqing gave the<br />
Distinguished Lecture<br />
Series in Statistical<br />
Science at the <strong>Fields</strong><br />
<strong>Institute</strong>. His two<br />
lectures, titled Vastdimensionality<br />
and<br />
sparsity and ISIS:<br />
A vehicle for the<br />
universe <strong>of</strong> sparsity,<br />
focused primarily on<br />
high-dimensional statistical modelling and feature selection.<br />
These aspects became important with the advent <strong>of</strong> mass data<br />
collection, advances in computation, and the discovery <strong>of</strong> new<br />
interplay between various natural and social sciences. In his public<br />
lecture, Fan outlined the problem <strong>of</strong> high dimensionality in fields<br />
Coxeter Lecture Series<br />
THE 2010 SUMMER THEMATIC PROGRAM ON THE<br />
Mathematics <strong>of</strong> Drug Resistance in Infectious Diseases was held at<br />
the <strong>Fields</strong> <strong>Institute</strong> during July and August. In association with<br />
this thematic program, Pr<strong>of</strong>essor Neil Ferguson was invited to<br />
the <strong>Fields</strong> <strong>Institute</strong> to deliver the Coxeter Lecture Series on<br />
Mathematical modelling <strong>of</strong> emerging infectious disease epidemics and their<br />
control.<br />
Ferguson, a Pr<strong>of</strong>essor <strong>of</strong> Mathematical Biology in the<br />
Division <strong>of</strong> Epidemiology, Public Health, and Primary Care <strong>of</strong> the<br />
Medical School at Imperial College, is a world leader in the use <strong>of</strong><br />
mathematical models in infectious disease epidemiology. He is the<br />
Director <strong>of</strong> the MRC Centre for Outbreak Analysis and Modelling.<br />
In the first lecture, Ferguson reviewed the development <strong>of</strong><br />
outbreak modelling over the last two decades and discussed the<br />
drivers which lead to more complex computational simulations<br />
being increasingly used replacing simpler compartmental models <strong>of</strong><br />
disease transmission. The second lecture discussed ways in which<br />
modelling can be optimally used to assist public health policymakers<br />
in their planning for and reaction to emerging infectious disease<br />
threats—an issue on which Ferguson is an expert, and which was<br />
14 FIELDSNOTES | FIELDS INSTITUTE Research in Mathematical Sciences<br />
Vast-dimensionality and Sparsity<br />
as diverse as bioinformatics, genetics, physics, and economics,<br />
discussing the twin challenges <strong>of</strong> noise accumulation and<br />
spurious correlations. The notion <strong>of</strong> sparsity or, more generally,<br />
homogeneity, was methodically described for feasible inference<br />
to high-dimensional problems. After an overview <strong>of</strong> the penalized<br />
likelihood approach for variable selection, Fan communicated the<br />
idea <strong>of</strong> large-scale screening and moderate-scale selection together<br />
with conditional inference as an effective solution to highdimensional<br />
problems. Using this approach in an analysis <strong>of</strong> U.S.<br />
housing price indices over a 30-year period, Fan demonstrated<br />
impressive prediction improvements.<br />
His second presentation was meant for a more specialized<br />
audience. Exploiting sparsity, Fan outlined a unified framework<br />
for solving high-dimensional variable selection problems. He<br />
applied iterative vast-scale screening followed by moderate-scale<br />
variable selection, resulting in a process called ISIS. Applying this<br />
process to multiple regression, generalized linear models, survival<br />
analysis, and machine learning, Fan demonstrated its overall<br />
reach via marginal variable screening and penalized likelihood<br />
methods. With tailored simulation studies and manipulation <strong>of</strong><br />
empirical data from disease classifications and survival analyses,<br />
Fan demonstrated the advantages <strong>of</strong> a folded-concave over convex<br />
penalty method on sure screening properties, false selection sizes<br />
and model selection consistency.<br />
Elif Fidan Acar (<strong>Toronto</strong>)<br />
Mathematical Modelling <strong>of</strong> Emerging Infectious Diseases<br />
<strong>of</strong> great interest to the thematic program participants. The third<br />
lecture focused on the potential impact <strong>of</strong> antiviral resistance<br />
during an influenza pandemic. He <strong>of</strong>fered several explanations<br />
for new findings that show the degree to which previous risk<br />
assessments concerning antiviral resistance in influenza pandemics<br />
have been over-pessimistic. In the lecture, Ferguson touched on<br />
the critical issue <strong>of</strong> the dependence <strong>of</strong> the final impact <strong>of</strong> resistance<br />
during a closed epidemic on the transmissibility <strong>of</strong> a sensitive<br />
and resistant virus, the mutation rate from one type to the other,<br />
and the level <strong>of</strong> seeding <strong>of</strong> both viral types at the beginning <strong>of</strong><br />
the epidemic. He argued that resistance is not likely to entail a<br />
substantial reduction <strong>of</strong> effectiveness <strong>of</strong> antivirals during the start<br />
<strong>of</strong> a pandemic, but that intensive drug use in this phase can lead<br />
to a higher degree <strong>of</strong> resistance in later epidemics. His concluding<br />
remark that “simple models suggest antiviral resistance could be a<br />
major issue in the first wave <strong>of</strong> a new pandemic, but allowing for<br />
spatial heterogeneity reduces speed <strong>of</strong> resistance” strongly echoed<br />
the theme <strong>of</strong> transmission heterogeneity <strong>of</strong> the two-week block <strong>of</strong><br />
this entire thematic program on mathematics for drug resistance.<br />
Jianhong Wu (York)