Friday,
October 2
CBCC214, 11:30 am
(refreshments at 11:15 am)
Prof. Tapabrata Maiti
Department of Statistics & Probability
Michigan State University
Title: Variable
Selection for Highdimensional Spatial Regression
[Abstract] Spatial regression is an important predictive tool in many
scientific applications. When there are many available predictors, it is
natural to select meaningful predictors. Penalized approach is proven to be
effective for both variable selection and estimation simultaneously. We will
talk some recent developments in this aspects of
handling large spatial data.
Friday,
October 23
CBCC214, 11:30 am
(refreshments at 11:15 am)
Prof. Zuoheng A. Wang
Department of Biostatistics
Yale School of Public Health
Title: Mathematical
Modeling of Dendritic Cell Population Dynamics
[Abstract] Dendritic cells
(DCs) are crucial to shape body response to invaders including bacteria and
viruses, thus serving as promising targets in immunotherapies for cancer and
other diseases. Knowledge of DC dynamics is limited due to experimental
difficulties. Combining both mathematical and experimental approaches, we
characterized DC population dynamics at the steady state as well as under
immune perturbation. Furthermore, using a computerbased algorithm, we
quantified the intricate relationships among key parameters, such as cell
death, proliferation and replenishment, many of which cannot be directly
measured, for both steadystate (normal immune system) and
cytokinestimulated (representing immune perturbation) DC populations in the
spleen. We found, in comparison to the steady state, DC proliferation
increased 11fold and surpassed cell death through shortening cell cycle
duration by 80%, resulting in dramatic expansion of the splenic DC
population. This study enables numerical evaluation of DC dynamics, a
significant implication on design of DCtargeted treatments.
Friday,
November 6
CBCC214, 11:30 am
(refreshments at 11:15 am)
Prof. S. Rao Jammalamadaka
Department of Statistics & Applied
Probability
University of California, Santa Barbara
Title: On the
Robustness of Bayes Predictions in Linear Models
[Abstract] We consider the
prediction problem for linear regression models with elliptical or spherically
symmetric errors, a special case of which is the multivariate tdistribution
with heavy tails. It is shown that the Bayes prediction density under the
elliptical errors assumption is exactly the same as that for normally
distributed errors when the prior information is objective or in the
conjugate family. Thus assuming that the errors have a normal distribution
when the true distribution is indeed elliptical, will not lead to incorrect
predictive inferences. This extends some earlier work of Arnold Zellner and others.
Friday,
November 20
CBCC214, 11:30 am
(refreshments at 11:15 am)
Prof. Duchwan
Ryu
Department of Mathematics & Statistics
Northern Illinois University
Title: Bayesian
Multiple Inflated Poisson Regression for Infection Data
[Abstract] We propose a
multiple inflated Poisson regression to model count responses containing
excessive frequencies at more than one nonnegative integer values in the
presence of covariates. The wellknown zeroinflated Poisson regression
combines binary regression and Poisson regression for an excess of zero
responses. To handle multiple excessive count responses we generalize the
zeroinflated Poisson regression by replacing the binary regression to the
multinomial regression. We discuss the properties of multiple inflated
Poisson model along with regression models when some covariates are
available, and use Bayesian computations for the complicated model
estimations. As an application, in the study of infectious diseases which
remain one of the greatest threats to human health, we observe excessive
zeros and ones in the number of changes of infections and utilize the
multiple inflated Poisson regressions.

Friday, March
11, 2016
CBCC138, 11:30 am
(refreshments at 11:15 am)
Prof. Hon Keung Tony Ng
Department of Statistical Sciences
Southern Methodist University
Title: Recent Advances
in Precedencetype Tests and Applications
[Abstract] In this talk,
we will provide a comprehensive overview of theoretical and applied
approaches to a variety of problems in which precedencetype test procedures
can be used. Then, some recent advances on precedencetype tests, including
the extension to progressively censored data, a twostage test for stochastic
ordering in two samples, and a sequential procedure for twosample problem
will be discussed.
Friday,
March 18, 2016
CBCC138, 11:30 am
(refreshments at 11:15 am)
Prof.
Jongphil Kim
Department of Biostatistics &
Bioinformatics
H. Lee Moffitt Cancer Center & Research Institute
and
Department of Oncologic Sciences
University of South Florida
Title: Computation of Bivariate
Normal and t probabilities, with
Application to Comparisons of Normal Means in Cancer Studies
(Changed to: A Novel Method for
Comparisons of Three Normal Means)
[Abstract] A novel method for the computation of the
bivariate normal and t probability is presented. With suitable
transformations, the probability over sets can be easily computed using exact
onedimensional numerical integration. An important application includes
computing the exact critical points for the comparisons of three normal means
for either the known or unknown variance problem. The critical points by
onedimensional integration can be computed using elementary numerical
methods and are more accurate than those by the approximation methods and
twodimensional integration methods. The comparisons of reliability
measurements from three populations are presented as an example of a known
variance case.
[Cancelled]
Friday, April 22, 2015
CBCC138, 11:30 am
(refreshments at 11:15 am)
Prof. Madhuri Mulekar
Department of Mathematics & Statistics
University of Southern Alabama
Title: Population Selection based on Count Data
[Abstract] The count data are popularly described using binomial,
Poisson and negative binomial distributions in managerial applications to
study consumer behavior, in entomological and ecological applications to
study species' behavior, and in quality control applications to study
systems' behavior. For each of these distributions, several solutions under
different configurations are available for the problem of selecting a best
population among the k available populations. Fixedsamplesize
procedures with the indifference zone approach will be discussed. Interesting
differences and similarities exhibited by these three distributions in the
existence of selection procedures and the consistency of selection procedures
will be explored.
Friday,
May 6, 2015
CBCC138, 11:30 am
(refreshments at 11:15 am)
Prof. Pankaj K. Choudhary
Department of Mathematical Sciences
University of Texas, Dallas
Title: Tolerance Bands
for Functional Data
[Abstract] Often the
object of inference in biomedical applications is a range that brackets a
given fraction of individual observations in a population. A classical
estimate of this range for univariate measurements
is a `tolerance interval.’ In this talk, we discuss its natural extension for
functional measurements, a `tolerance band,' and describe a methodology for
constructing its pointwise and simultaneous
versions that incorporates both sparse and dense functional data. Assuming
that the measurements are observed with noise, the methodology uses
functional principal component analysis in a mixed model framework to
represent the measurements and employs bootstrapping to approximate the
tolerance factors needed for the bands. The proposed bands also account for
uncertainty in the principal components decomposition. Simulations show that
the methodology has generally acceptable performance unless the data are
quite sparse and unbalanced, in which case the bands may be somewhat liberal.
The methodology is illustrated using two real datasets, a sparse dataset
involving CD4 cell counts and a dense dataset involving core body
temperatures.
