Department of Mathematics and Statistics Calendar
February 14 - 20, 2005
Monday, February 14, 2005
Tuesday, February 15, 2005
10:45 AM MATH 115 Instructors Meeting 400 MSc
Wednesday, February 16, 2005
10:30 AM Algebra Seminar 400 MSc
Dr. Corneliu HOFFMAN, BGSU
"Flips for Algebraic Groups, II"
11:30 AM Statistics Seminar **140 MSc** *note location change*
Dr. N. BALAKRISHNAN, Visiting Lukacs Professor
"Progressive Censoring V: Goodness-of-Fit Tests"
12:30 PM Calculus Seminar 459 MSc
TBA
3:30 PM Analysis Seminar 400 MSc
Dr. Juan BES, BGSU
"TBA"
**3:30 PM Statistics Candidate Colloquium** 459 MSc
"Robust Estimation via Measurement Error Model
Likelihoods with Applications to Location-Scale Models"
Thursday, February 17, 2005
**3:30 PM Statistics Candidate Colloquium** 459 MSc
"Diagnostic Measures and Selection Criteria for
Mixed Models Based on Kullback-Leibler Information"
ABSTRACT: This talk consists of two parts. In the first part, a
diagnostic defined in terms of the Kullback-Leibler directed divergence
is developed for identifying cases which impact the prediction of the
random effects in a mixed model. The diagnostic compares two
conditional densities governing the prediction of the random effects:
one based on
parameter estimates computed using the full data set; the other based
on parameter estimates computed using a case-deleted data set. We
present
the definition of the diagnostic and a computational formula for its
evaluation. We illustrate the effectiveness of the measure using a
simulated data set. The performance of the measure is also investigated
in an application where exam scores are modeled using a mixed model
containing a fixed exam effect and a random subject effect.
In the second part, we extend the results of Shibata (1997) to
propose five bootstrap-corrected variants of AIC, denoted by AICb1-AICb5,
for the purpose of small-sample mixed model selection. These variants
are asymptotically equivalent, and provide asymptotically unbiased
estimators
of the expected Kullback-Leibler discrepancy between the true model and a
fitted candidate model. The performance of the criteria is investigated in
a simulation study where the random effects and the errors for the true
model are generated from a Gaussian distribution. Parametric,
semiparametric, and nonparametric bootstrapping are employed. Our
results indicate that under parametric bootstrapping, the variants
exhibit the best overall bias properties. For correctly specified or
overfit models, the criterion originally introduced by Efron (1983,
1986), AICb1, provides a
less biased estimator of the expected discrepancy than AIC and the
other proposed variants. The criterion originally introduced by
Cavanaugh and
Shumway (1997), AICb2, exhibits a higher success rate in identifying the
correct model than AIC and the other variants. Overall, the bootstrap
AIC variants appear to serve as effective tools for selecting a mixed
model of appropriate dimension.
Friday, February 18, 2005
**12:00 PM Calendar Information due to Cyndi
for inclusion in next week's calendar listing**
**1:30 PM Director of Service Mathematics Candidate Colloquium**
459 MSc
"Wavelets, Multiscale Analysis and Their Applications"
ABSTRACT: We begin with a brief overview of wavelet analysis. We then
compare wavelet transforms with Fourier transforms and state their
properties. Multiresolution analysis provides a natural framework of
the construction of wavelets. We describe its insight by an
illustration from a coarse approximation to a higher resolution
approximation. The computational efficiency and accuracy of the
multiscale methods have allowed us to make improvements in many
applications. Several wavelet based multiscale computational methods
will be introduced. Applications on image processing
and numerical solutions of partial differential equations will be mentioned.
Saturday, February 19, 2005
Sunday, February 20, 2005
A list of mathematics seminars by subject and other seminars at BGSU is available here.
If you have comments or material for the calendar, send e-mail to Cyndi
Patterson,
If you wish to be placed on the e-mail
distribution list, send e-mail
to Craig Zirbel,
Previous calendars are available individually
or in one single file for searching.
Return to Math & Stat Home Page / BGSU Welcome Page
/
/ Disclaimer