| Seminars | |||
|
Seminars 2006-2007
Wednesday, October 4, 2006 Friday,
October 6, 2006, 3-4 pm (sponsored jointly with Mathematics Dept) "Adaptive Coherent Interferometric Imaging in Random Media and Optimal Waveform Design" (work incollaboration with George Papanicolaou (Stanford) and Chrysoula Tsogka (U. Chicago) I
will discuss a robust, coherent interferometric approach for imaging
of strong reflectors in cluttered Depending
on the strength of the inhomogeneities and the distance of propagation,
the effect of the clutter on the wave field can be classified as: The
coherent interferometric approach discussed in this talk addresses
the second case and it can be viewed as a statistically smoothed migration
technique that exploits systematically the spatial and I
will describe in some detail the method, its statistical stability
and resolution. In particular, I will explainhow the suppression of
the clutter noise in the data requires a certain type of smoothing Finally, I will describe some recent results on the question of optimal waveform design for imaging with arrays, in both cluttered and smooth (deterministic) media.
South
Hall 5607F Finance models with a finite number of states and dates have a number of properties that may or may not extend to their infinite counterparts. Whether they do depends how equilibrium is defined. Under sequential equilibrium many properties do not extend to infinite settings. We propose an alternative equilibrium concept that is closer to classical finite Walrasian equilibrium. This involves appending a date called \uffff~H~^ to the finite dates and defining a topology such that the payoff index set so expanded is compact. Payoffs of infinite portfolio strategies are defined as limits of payoffs of finite portfoliostrategies. This setup allows a simplified mathematical treatment of a number of topics that are unwieldy when modeled in a setting where the payoff index set is not compact. Topics discussed include Ponzi schemes, payoff bubbles and the doubling strategy. Wednesday,
November 1, 2006 A
flexible class of prior distributions for the covariance matrix
of a multivariate normal distribution will be discussed. Approximate
and exact posterior moments for the parameters of interest can be
calculated via a Metropolis-Hastings/MCMC algothrithm. A subset
of the Project Talent Wednesday,
November 8, 2006 A general method will be presented for deriving the limiting behavior of estimators that are defined as the values of parameters optimizing an empirical criterion function. The asymptotic behavior of such estimators is typically deduced from uniform limit theorems for rescaled and reparametrized criterion functions. The new method can handle cases where the standard approach does not yield the complete limiting behavior of the estimator. The asymptotic analysis depends on a decomposition of criterion functions into sums of components with different rescalings. The method will be explained by examples from shorth estimation, k-means clustering and partial linear models
Wednesday, November 15, 2006 In this talk, we explore issues to analyze functional data (often termed longitudinal data when it is sparse) which arise in medical studies. These functional/longitudinal data serve as biomarkers for disease progression and are often not observable after an event, such as death, occurs. This triggers informative dropout. Since the functional/longitudinal data are related to the event-time, marginal approaches to model the functional/longitudinal processes will induce bias, and one way to remove the bias is to model both the event and longitudinal processes simultaneously. Such an approach is termed joint modeling of longitudinal and survival data in the literature. We will explore several intriguing and challenging issues in joint modeling. Typically, a parametric longitudinal model is assumed to facilitate the likelihood approach. However, the choice of a proper parametric model turns out more illusive than in standard longitudinal modeling where no survival end-point is considered. Furthermore, the computational burden and stability are important concerns in the joint modeling setting. To deal with these challenges, we propose a simple semiparametric random effects model for the functional/longitudinal data and illustrate this through numerical studies and data analysis. Another challenge in the joint modeling approach is the high dimensionality problem involved in the EM-algorithm. We will show how the method of sieves helps to resolve this difficulty and discuss the asymptotic properties of the proposed sieve estimators. Wednesday,
November 29, 2006
South Hall 5607F 3:15 PM, Refreshments served at 3:00 PM Li Qin, Fred Hutchinson Cancer Research Center, Seattle, Washington Assessing Surrogate Endpoints in Vaccine Trials with Case-Cohort Sampling and the Cox Model Assessing
immune responses to study vaccines as surrogates of protection (SoPs)
plays a central role in vaccine clinical trials. We consider such
surrogate endpoint assessment in a randomized placebo-controlled
trial with case-cohort sampling of immune responses and a time to
event endpoint. We extend the principal surrogate definition under
the principal strata framework proposed by Frangakis and Rubin (2002)
and Gilbert and Hudgens (2006), and introduce causal estimands that
measure the value of an immune This is joint work with Peter Gilbert, Dean Follmann and Dongfeng Li.
Friday,
Jan 12, 2007 Alexander
Schied, Berlin University of Technology Due to the complexity of financial price processes, their mathematical models are often subject to model misspecification. In this talk we present some recent results on the robustness of certain trading strategies with respect to model uncertainty. In the first part, we consider the robustness of the Delta hedging strategy of an exotic derivative with respect to realized volatility when the underlying model is a local volatility model. Our analysis is based on volatility comparison techniques for SDEs. In the second part, we focus on the construction of optimal investment strategies for an investor who is averse against both risk and model uncertainty. Here one can use or combine several techniques including convex duality, nonlinear PDEs, and robust statistical test theory. In some special cases, the problems considered in parts one and two are closely related to each other.
Wednesday, February 21, 2007 How many knots are enough? This question addresses the problem of estimating the number of distinct topological knot types and their proportion in the space of (equilateral) polygonal knots with a fixed number of edges. For very small numbers of edges, one knows the number of knot types and can estimate their proportion but, for largernumbers of edges, only rough estimates are available. Estimates derive from Monte Carlo explorations of the (equilateral) polygonal knot space and an analysis using the HOMFLY polynomial as a surrogate for the topological knot type. As a consequence, one is interested in knowing how large a sample of knots is needed to give a good estimate of the number of topological knot types as reflected by distinct HOMFLY polynomials. Some theoretical and experimental efforts concerning this question will be discussed. Feb
28, 2007
March 14, 2007
Applying Model Reference Adaptive Search to American-style Option Pricing Abstract
This paper considers the application of stochastic optimization
methods to American-style option Wednesday,
April 25, 2007 The
Skorokhod embedding problem amounts to stopping a Brownian motion
to hit a target density; it has interesting implications for finance:
Dr. Bruno Dupire, Bloomberg and NYU May
23, 2007 - SOBEL LECTURE Micronutrient
Supplementation, Birth Weight and Infant Mortality; On Estimation
of Percentile-Specific, Mediated Intervention Effects Thus, two key scientific questions are whether the effect of the treatment on survival varies across the birth weight distribution and whether the effect of the treatment on survival is mediated wholly or in part by increases in birthweight. This talk will define and estimate population- and subject-specific parameters that describe the treatment effects on birth weight and on survival as functions of the percentiles of the birth weight distribution. Sensitivity of some subject-specific inferences to unverifiable assumptions will be demonstrated.
May
30, 2007 Sanjib Basu, Northern Illinois University Bayesian competing risks analysis of cancer survival data from the SEER program The rates of cancers, including age-adjusted mortality and incidence rates, depict a general increase over the last 30 years. These led some to question the success of the war on cancer. The rates of many other competing diseases, on the other hand, have declined. It has been hypothesized that this decline is somewhat responsible for the rise in cancer rates. We consider competing risks analysis of cancer survival data that considers the simultaneous risks of cancer as well as other causes. The cure rate survival models for cancer postulates a fraction of the patients to be cured from cancer. We propose a model that incorporates competing risks and, at the same time, allows a fraction of patients to be cured. We describe Bayesian analysis of this model, discuss both conceptual and methodological issues related to model building and model selection, and consider application in survival data for breast and prostate cancer patients in the SEER registries of the National Cancer Institute (NCI). June
4 ,
2007 (Monday) "NCI’s Biostatistics Grant Portfolio and NIH Funding Mechanism” The talk consists of two parts. In Part I, I will talk about our website: www.statfund.cancer.gov. This website contains information about a large proportion of the NIH funded grants in biostatistical methods in areas of application to cancer epidemiology, treatment, survival and prevention. These grants are housed in the Division of Cancer Control and Population Sciences (DCCPS) at the National Cancer Institute (NCI). I will also discuss various funding opportunities in statistics at the NCI. In Part II, I will go over the NIH funding mechanisms and discuss the NIH grant review process in great detail.
|
|||