Some talks

Most of these talks are based upon work supported by the National Science Foundation under Grants: DMS-1521145, DMS-1407397, DMS-0906056, DMS-0604939, DMS-0306612, and DMS-0072445. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The reports are here.

Relatively recent talks:

Noether Senior Scholar Award Talk
JSM 2020
Empirical likelihood: some history and new directions
Square root rule for adaptive importance sampling
Bayescomp 2020, University of Florida
Safe and simple way to combine results from multiple iterates.
Six percent power and barely selective inference
WHOA-PSI 4, Washington University of St. Louis
How to cope with sign error issues raised by Andrew Gelman.
Variable importance in statistics and real life
2019 Bradley Lecture: after dinner, April 2018
Ideas about variable importance that I've been kicking around for about 10 years.
Tuning the tie-breaker design
Stanford statistics seminar, October 2018
Joint with Hal Varian (Google)
London Mathematical Society and CRISM Summer School in Computational Statistics
July 2018, University of Warwick web page

Foundations of Monte Carlo, short course

  1. Monte Carlo, what, why and how
  2. Random vectors
  3. Variance reduction
  4. Importance sampling
  5. Introduction to MCMC

Quasi-Monte Carlo, short course

  1. Survey of QMC
  2. QMC for MCMC
  3. QMC outside the unit cube
  4. Stolarsky invariance and gene set testing

Statistically efficient thinning of a Markov chain sampler
Conference for Charlie Geyer, April 2018, University of Minnesota, Minneapolis

Important sampling the union of rare events, with an application to power systems analysis
SAMSI workshop, December 2017, Duke University
(The bonus material on thinning to improve statistical efficiency of MCMC was not presented here. It was shown at the Geyer conference; slides above.)
Quasi-Monte Carlo, beyond the unit cube
Monte Carlo Methods 2017, Montreal, July 2017

Bayesian Empirical Likelihood
11th Conference on Bayesian Nonparametrics, BNP 11, Paris, June 2017

Shapley value for measuring importance of dependent inputs
Statistical Perspectives on Uncertainty Quantification, GA Tech, May 2017

Method of moments for large crossed linear mixed models
University of Chicago, UC Davis, Stanford, 2017

Adaptive Importance Sampling
Grid Science 2017, Santa Fe NM, January 2017

Permutation p-value approximation via generalized Stolarsky invariance
International Society for Non-parametric Statistics
ISNPS 3, Avignon France, June 2016
I missed this meeting due to an Air France strike. These are the slides I would have given, with annotations for some things I would have said aloud.
ANOVA and Sobol' indices tutorial: UQ 2016
Lausanne Switzerland, April 2016
QMC Tutorial from MCMSki 5
Lenzerheide Switzerland, January 2016
Picking k in factor analysis
With J. Wang
Bi-cross-validation slides
Moment estimation for large unbalanced crossed random effects
With K. Gao
Variance of estimated variance components slides
25th Eurographics Symposium on Rendering
Keynote on sampling: slides
Self-concordance for empirical likelihood
Canadian Journal of Statistics award 2014: talk| en Français
Empirical likelihood
Talk in honour of David Sprott
ANOVA, global sensitivity, Sobol' indices and all that
A tutorial from MCQMC 2014 in Leuven
Data enriched linear regression
With A. Chen and M. Shi
JSM slides | U. Michigan slides
Generalized Sobol' indices
Stanford dept talk | Notation crib sheet
Balanced adjusted empirical likelihood
First part based on work with S. Emerson
The rest is unpublished notes on EL for regression
ICSA slides
MCQMC talks, Sydney
With D. Eckles and S. Emerson
First 21 slides cover unpublished random projection work with S. Emerson on testing means when d >> n
The rest is crossed effects bootstrap with D. Eckles
covering basic Monte Carlo
Quasi-Monte Carlo for Markov chain Monte Carlo
Based on work with S. Chen, J. Dick, M. Matsumoto, T. Nishimura
IMS Asia Pacific Rim version| MCQMC version
Pearson's meta-analysis: wrongly thought inadmissible for ~50 years!
Based on a problem arising in the S. Kim dev bio lab
Wharton slides
Visualizing networks and other bivariate heavy tailed data
With J. Dyer
NIPS slides
Bootstrapping sparse crossed random effects data (eg Netflix or Facebook)
With D. Eckles
Google Stat Foo| Iowa State U. Snedecor Lecture
Semi-supervised learning on graphs
Based on Y. Xu's dissertation, and J. Dyer
ASA SF Chapter
Cross-validation of SVD, NMF and other outer product matrix data models
With P. Perry
JSM version
Infinitely imbalanced logistic regression Few 1s, many 0s
U FL winter workshop
Crossed two way random effects bootstrap (eg Netflix)
Latent variables in the AGEMAP data
With Kim lab and P. Perry
NIA slides
Robust hybrid of Ridge and Lasso (via Berhu penalty)
Snowbird slides
Empirical likelihood survey Fields Seminar
QMC and RQMC for unbounded integrands
MCQMC 2004
Estimating mean dimensionality
With R. Liu
MCQMC 2004

Older talks:

Some of these were originally rendered in postscript. They have been converted to PDF but not all PDF viewers like the result.
Plaid models
With L. Lazzeroni
A gene recommender for C. elegans
With A. Villeneuve Lab and S. Kim Lab
Bay area worm meeting| World worm meeting| Synmuv phenotypes| Actual C elegans in action
Quasi-regression for black box functions
With T. Jiang
Quasi-Monte Carlo with control variates
With F. Hickernell and C. Lemieux
Latin supercube sampling