Interests include development of Bayesian methods for problems in Astronomy, and in High-Energy Physics. I have been heavily involved in a project for studying stellar evolution and star formation histories using Color-Magnitude diagrams. Using photometric data it is possible, by comparison with the output of complex models derived from physical properties, to estimate the age (and other properties) of both individual stars and clusters of stars. The highly structured nature of the observations and underlying physical processes leads to some extremely interesting computational challenges. Another project I have worked is the so-called Banff challenge that I and a couple of other Harvard graduate students worked on.
With the development of increasingly respected (and scrutinized) global and regional climate models (GCM/RCMs), future projections of climate conditions on both a local and continental scale are of importance to both policy makers and the wider public. RCMs and GCMs are typically based on fluid dynamics, and involving the solution of partial differential equations. As such, RCM/GCMs are inherently deterministic models, leaving a role for statistics in better quantifying the uncertainties involved. My research aims to coherently combine both RCMs and GCMs with point-source data from the US Historical Climatology Network (USHCN) to improve the spatial-resolution of climate projections.
There are an icreasing number of applications involving, and an increasing body of theory behind, the replacement of full/genuine likelihoods by various forms of artifical likelihood (including profile likelihood, marginal likelihood, pseudo-likelihood, partially-specified likelihood, and estimating equations). How/can we combine these forms of 'likelihood' with prior beliefs to make coherent Bayesian inference?
Bayesian methods for fitting Generalized Linear Mixed Models are often application-specific and widespread access to reliable and flexible code to fit them is hard to find. My interests are primarily in the methodology and computation; and we are currently developing a general-purpose methodology for computation in GLMMs using an Ancillarity-Sufficiency Interweaving Strategy (ASIS).
Probability Matching Priors are essentially a broad class of prior distributions designed to provide Frequentist validity of Bayesian posterior credible intervals (up to some order ofapproximation). Despite being theoretically appealling to many they are seldom used outside of the simplest cases because of the computational obstacles involved.
Slides and a videotaping of my Phystat 2007 talk on "Probability Matching Priors in LHC Physics" are available here.
JSM 2007 slides from a different version of the talk are available here.
Baines, P.D., Zezas, Meng, X.-L., Lee, H. (2010) "Interwoven EM Algorithms" (in preparation)
Baines, P.D., Meng X.L., Hayhoe, K. (2010) "Spatiotemporal Downscaling for Local Climate Prediction" (in preparation)
Baines, P.D., Zezas, A., Kashyap, V.K., Meng, X.-L., Lee, H. (2009) "Bayesian inference for ages of stellar populations from multi-band color-magnitude diagrams" (in preparation)
Baines, P.D., Meng, X.-L. (2007) Probability Matching Priors in LHC Physics, Proceedings of Phystat 2007, CERN Yellow Book
PhD Thesis Defense, Department of Statistics, Harvard University, April 30th 2010; "Interwoven EM Algorithms"
Statistical Frontiers of Astrophysics, IPMU, University of Tokyo, Sept 28th-Oct 2nd 2009; "Bayesian analysis of stellar populations"
Joint Statistical Meetings, Washington DC, Aug 1st-6th 2009; "Spatio-Temporal Downscaling for Local Climate Prediction"
213th American Astronomical Society Meeting, Long Beach, CA, January 4th-8th 2009; "Ages of stellar populations from color-magnitude diagrams" (Invited speaker)
The 7th World Congress in Probability and Statistics, Singapore, July 14-19, 2008; "Constructing length and coverage-based prior distributions" (Laha travel awardee)
Joint Statistical Meetings, Salt Lake City, July 29th-Aug 2nd; "Upper Limits for Source Detection in the Three-Poisson Model"
PHYSTAT LHC Workshop on Statistical Issues for LHC Physics, CERN, Geneva, Switzerland, 27-29 June 2007; "Probability Matching Priors in LHC Physics" (Full proceedings are available here)