Bayesian Methods (David van Dyk, chair)
Miguel Dumett (USC Department of Mathematics)
Data Analysis for the Simulation of Alcohol Concentration
Thursday 10:30-10:50, San Rafael
Abstract:
A model for the transport of alcohol from blood throughout the skin and to the skin surface is proposed. The parameter models are calibrated to a subject by a least squares procedure that utilizes data coming from a breath alcohol device and a dermal sensor. Given the parameter estimates it is possible to simulate accurately skin vapor alcohol concentration and apply the Kalman filtering approach for normal errors to update predictions of alcohol concentration in real time. A generalization of the Kalman filtering methodology to non-normal errors is introduced to predict the evolution of alcohol concentration.
Byron Ellis (Harvard University)
Wing Wong (Stanford University)
A Fast Order-directed Bayesian Network Sampler With Applications to High Throughput Genomics and Proteomics
Thursday 10:50-11:10, San Rafael
Abstract:
Sampling Bayesian Network structures is a difficult MCMC task fraught with problems including computational burden and slow mixing and local trapping. To obtain samples efficiently we have developed a two-phase sampling scheme that combines equi-energy sampling techniques with an efficient scoring calculation to construct a sampler that has performed well in real data analysis settings, particularly those related to high-throughput biology experiments.
A. Ptak (Johns Hopkins University)
Bayesian Parameter Estimation for X-ray Astrophysics Observations with Disparate Data Sets
Thursday 11:10-11:30, San Rafael
Abstract:
X-ray satellites often contain multiple telescopes that are operated
simultaneously, resulting in several data sets for a given target. The
instrumental response and background characteristics of the detectors
often differ which complicates the interpretation of the results. A similar situation occurs when astronomers wish to analyze observations of
a source performed with different telescopes that have different performance characteristics (usually different spatial and spectral resolution). We
have developed code to address the specific problem of limits on the
luminosity of a source that was not detected when observed with multiple
telescopes. We will discuss the generalization of this effort to include
Bayesian parameter estimation for spectro-spatial analysis of multiple
X-ray observations. Specific examples will include determining limits on
faint spectral emission lines and the analysis of observations of galaxies
which often contain variable point sources (black holes and neutron star
X-ray binaries) intermixed with diffuse emission (gas at a temperature of
~ 10 million degrees). In the latter example, observations with poor
spatial resolution do not resolve many of the point sources from the
diffuse emission.
Sarjinder Singh (Department of Statistics, St. Cloud State University)
Calibrated Empirical Likelihood Estimation Using a Displacement Function: Sir R.A. Fisher's Honest Balance
Thursday 11:30-11:50, San Rafael
Abstract:
In this paper, we propose a new method to calibrate the estimator of the general parameter of interest in survey sampling. We demonstrate that the linear regression estimator due to Hansen, Hurwitz and Madow (1953) is a special case of this. We reconfirm that the sum of calibrated weights has to be set equal to sum of the design weights within a given sample as shown in Singh (2003, 2004, 2006) and Stearns and Singh (2005). Thus, it shows that the Sir R.A. FisherŐs brilliant idea of keeping sum of observed frequencies equal to that of expected frequencies leads to a "Honest-Balance" while weighing design weights in survey sampling. The major benefit of the proposed new estimator is that it never fails like the pseudo empirical likelihood estimators listed in Owen (2001). The main endeavor of this paper is to bring a change in the existing calibration technology, which is based on only positive distance functions, with a displacement function that has the flexibility of taking positive, negative, or zero value. At the end, the proposed technology has been compared with its competitors under several kinds of linear and non-linear non-parametric models using an extensive simulation study. This paper will encourage a lot of scientists and researchers to think more on these lines.