My research interests include Bayesian modeling and computation, astrostatistics, machine learning methods, and measuring statistical information especially in the context of experimental design. I have been a member of the international CHASC Astrostatistics collaboration since 2010. The best way to contact me is through email at djones@stat.tamu.edu.
Some of my focuses are:
-
Developing interpretable measures of statistical information for classification, model selection, and hypothesis testing problems. The information we need differs by context and I am interested in the information overlap and separation when considering several inference goals. I apply this to the problem of scheduling follow-up telescope observations in order to better classify astronomical lightcurves (time series).
-
Quantifying the impact of prior information.
-
Methodology and theory for constructing efficient Monte Carlo estimators of normalizing constants (or integrals more generally), especially in the context of multi-modal densities. I also apply this methodology to exoplanet detection.
-
Scalable Bayesian clustering and its application to astronomy.
-
Statistical and machine learning methodology for exoplanet detection in the presence of stellar activity. This is essentially a signal detection problem where the data are a high dimensional time series and the noise is highly structured and non-stationary.
-
Bayesian hierarchical modeling and inference for images in astronomy and neuroscience. Goals include faint object detection, overlapping object separation, and characterization of transient phenomena.
For a complete list of publications please refer to my Google Scholar profile.