Bayesian Nonparametrics: Synergies between Statistics, Probability and Mathematics (BNP-SSPM): June 29-July 2, 2015


The workshop was held at SAMSI, Research Triangle Park, NC.


Bayesian Nonparametrics (BNP) is a rapidly evolving area at the intersection of statistics, machine learning, probability and computer science. The focus is on modeling infinite-dimensional unknown objects that may consist of curves, surfaces, processes or probability measures. There are currently several vibrant communities focused on different aspects of BNP ranging from design of new processes motivated by machine learning and high-dimensional data problems, asymptotic properties, probabilistic properties, computation and intersections of these different areas. The area is amazingly multi-disciplinary, with leading researchers having diverse backgrounds in mathematics, statistics, probability theory, computer science and electrical engineering. The goal of this four-day workshop is to bring together a group of leading researchers having different perspectives on BNP including “outsiders” working on related areas relevant to BNP, such as optimization and probability, with the goal of spurring new collaborative projects aimed at developing transformative new approaches and high impact scientific tools.

The program followed and involved a subset of participants from the 10th Conference on Bayesian Nonparametrics that was held in Raleigh, June 22-26, 2015. However, the summer program was a stand alone event, with its focus squarely placed on the synergy between mathematics, computer science, probability and statistics relevant to Bayesian Nonparametrics and involved its own set of leading researchers in the best interest of the cause. A primary aim of the program was to facilitate interaction between leading experts and graduate students and young researchers to help train the next generation BNP researchers.

Questions: email

Schedule and Supporting Media

Speakers, Titles, Abstracts

Monday, June 29, 2015

Welcome and Introduction Sujit Ghosh, SAMSI
David Dunson, Duke University
Session I: Overview of Bayesian Nonparametrics
A Gentle Introduction to Nonparametric Bayes
Steve MacEachern, Ohio State University  
Session 2:
Panel Discussion on Multiresolution Methods
Antonio Canale, University of Turin
Li Ma, Duke University
Session 3: Breakout Groups on Multiresolution Methods
(SAMSI Rooms: 150, 259, 219, 203, NISS Room 104)
Session 4: Recap and Discussions of the Breakout Groups

Tuesday, June 30, 2015

(SAMSI Rooms: 150, 259, 219, 203, NISS Room 104)

Session 5: Research Seminars
Linear Response Methods for Accurate Covariance Estimates from Mean Field Variational Bayes
Confidence in Bayesian Uncertainty Quantification?
Tamara Broderick, MIT
Botond Szabo, CREST
Session 6:
Panel Discussion on High-dimensional Analysis
Surya Tokdar, Duke University
Peter Orbanz, Columbia University
Session 7: Breakout Groups on High-dimensional Analysis
Session 8: Recap and Discussions of the Breakout Groups

Wednesday, July 1, 2015

Session 9: Research Seminars
Scalable Bayes via Fast Computation of Barycenter of Subset Posteriors 
Posterior Concentration in Gaussian Process Regression using Wasserstein BvMs
Sanvesh Srivastava, Duke University
Anirban Bhattacharya, Texas A&M University
Session 10: Panel Discussion on Statistical Emulation and Optimization Moderators:
Ralph Smith, North Carolina State University
Robert Wolpert, Duke University
Session 11: Breakout Groups on Statistical Emulation and Optimization
(SAMSI Rooms: 150, 259, 219, 203, NISS Room 104)
Session 12: Recap and Discussions of the Breakout Groups

Thursday, July 2, 2015

Session 13: Panel Discussion on Theoretical Developments Moderators:
Natesh Pillai, Harvard University
Subhashis Ghosal, North Carolina State University
Session 14: Breakout Groups on Theoretical Developments
(SAMSI Rooms: 150, 259, 219, 203, NISS Room 104)
Concluding Remarks