Matthew T. Pratola

John Yannotty's new work on Bayesian Model Mixing (BMM) has been posted to arXiv! In this research John proposes a smooth BART-based emulator for learning the latent BMM weight functions using an elegant prior construction that enables matrix-free updates. His insight balances between the smoothing effect of probabilistic splits and the localization effect of trees, and he derives a semivariogram approach for calibrating the prior. The model's capability is demonstrated by mixing Global Climate Models (GCMs), where he further proposes posterior projection techniques for interpreting the space of GCMs being mixed.

John Yannotty and his BAND collaborators Kevin Ingles, Dan Liyanage and Alexandra Semposki have published a paper on their Bayesian Model Mixing software package Taweret in the Journal of Open Source Software!

Version 0.3 of the BAND Framework, including support for Bayesian Model Mixing, has been released! Explore it here.

John Yannotty's first paper on Bayesian model mixing has been published at Technometrics! The code is also available at John's repo and has been integrated into the Taweret package in the Bayesian Analysis of Nuclear Dynamics (BAND) software framework, complete with code and example Jupyter notebooks, at the BAND repo.

Our latest BAND paper, Model Mixing Using Bayesian Additive Regression Trees, has been posted to arXiv! In this work, John Yannotty develops a model for mixing multiple computer simulators, in this case EFT expansion models of nuclear dynamics from our physics collaborators.

Vincent Geel's paper on MCMC for discrete spaces has been published at the Journal of Statistical Computation and Simulation! Vincent's so-called taxicab sampler builds on ideas from the Hamming ball sampler, and is demonstrated on a discrete response tree model. Read it here.

We have a Postdoc position available to work on the BAND project! Exciting applications in nuclear physics using Bayesian model mixing techniques. Read more here.

Akira’s latest paper, “Using BART to Perform Pareto Optimization and Quantify its Uncertainties,” has been accepted to the special issue on Industry 4.0 at Technometrics! Joint work with Tom Santner and Ying Sun. Congrats Akira, 2 special issues in a row is pretty amazing!

Akira’s first paper, “Assessing variable activity for Bayesian regression trees,” has been accepted to the special issue on Sensitivity Analysis of Model Outputs at Reliability Engineering and System Safety. Joint work with Tom Santner. Congrats Akira, and have fun with the postdoc @Duke!

Our NSF grant “Bayesian Analysis of Nuclear Dynamics” has been awarded! This large, multi-institutional award seeks to bring the statistical tools of model mixing, calibration and uncertainty quantification to important problems at the frontier of nuclear physics. You can read the official announcement of this incredibly exciting project here and here.

Our latest BART paper, “Assessing variable activity for Bayesian regression trees,” has been posted to arXiv! In this work, Akira Horiguchi investigates alternatives to the usual split-count variable activity metrics. Building on the idea of Sobol indices, Akira finds that the popular split-count can, in fact, be quite misleading. Subsequently, an efficient algorithm for computing such indices in BART models is developed and an interesting connection to the usual split-count metric is described. We will be adding this feature to the OpenBT code base over the next few months! https://arxiv.org/abs/2005.13622

Our grant “Innovations for Bayesian Tree Ensemble Methodology” submitted to the National Science Foundation (NSF) has been awarded! This is a collaborative research project with Dr. Rob McCulloch (co-PI, Arizona State University), Dr. Ed George (co-PI, The University of Pennsylvania) and myself (PI, OSU). More exciting BART innovations incoming!

Our grant “An Advanced Spatio-temporal Statistical Methodology for Impact Studies on Air Quality and Renewable Energy” submitted to the prestigious King Abdullah University of Science and Technology (KAUST) Competitive Research Grant (CRG) program has been awarded! This is a collaborative research project with Dr. Ying Sun (PI, KAUST), Dr. Brian Reich (co-PI, NCSU) and myself (co-PI, OSU). Most excitingly, this grant will support some early stages of work for the dream project!