Students in the Master's programs in Statistics:
Questions about specific course work should be directed to the
advising team (email address: email@example.com) or to your assigned faculty advisor.
Please go to this page to this page
set up an appointment
with me or other advisors. Students in the dual Master's program are
welcome to sign up for an appointment with me.
- Office hours in Winter 2019: 10--12pm, Thursdays, WH 461
Prospective PhD students: Thank you for your interest. Admissions decision is
made a graduate admissions committee, please see
this link for further information. My apology if I am unable to respond to your enquiry
due to the large volume of such emails.
Synopsis: Statistical inference is the computational process of turning data into statistics, prediction
and understanding. I work with richly structured data, such as those
extracted from texts, images and other
- Nonparametric Bayesian statistics
- Machine learning and optimization
- Hierarchical, mixture and graphical models
- Spatiotemporal and functional data analysis
- Stochastic, variational and geometric methods in statistical inference
In recent years I have gravitated toward a
field in statistics known as Bayesian nonparametrics, which provides
a fertile and powerful mathematical framework for the development of
many computational and statistical modeling ideas. The spirit of
Bayesian nonparametric statistics is to enable the kind of
inferential procedures according to which both the statistical modeling
and computational complexity may adapt to increasingly large and
complex data patterns in a graceful and effective way.
In this framework, stochastic processes and random measures, along with
latent variable models such as mixture, hierarchical and graphical
models figure prominently. My group's research is centered around
between statistical inference and the theory of optimal transport
that arises in the learning of complex hierarchical models.
My motivation for all this came originally from an interest in
machine learning, which continues to be a major source of active
A primary focus in our machine learning research is to develop more effective
inference algorithms using variational, stochastic and geometric viewpoints.
Fun statistics: number of papers I have written that has a giant's name
in the title, as of twenty eighteen: Bayes = 6; Dirichlet = 4; Gauss = 2; Wasserstein = 1; Lyapunov = 1;
Names I wish to appear on my paper's title: Blackwell, De Finetti, Riemann, Stein,
Former students, postdocs and visitors
- Mikhail Yurochkin PhD 2018, Research scientist, IBM Research, Cambridge, MA
- Giuseppe Di Benedetto Visiting PhD student from Oxford University,
- Nhat Ho PhD 2017, joint with Ya'acov Ritov;
Postdoctoral fellow, University of California, Berkeley
- Hossein Keshavarz PhD 2017, joint with Clay Scott;
Postdoctoral fellow at the IMA, University of Minnesota
Federico Camerlenghi Postdoctoral visiting scholar, April--May 2016;
Assistant Professor, University of Milano-Bicocca
- Zhaoshi Meng PhD 2014, joint with Al Hero; Senior Researcher, Vicarious, CA
Arash Ali Amini Postdoctoral fellow 2011--2014; Assistant Professor, University of California, Los Angeles
- Vijay Manikandan Janakiraman PhD 2013, joint with Dennis Assanis; Research Scientist, NASA's Ames Research
- Jian Tang
Visiting PhD student from Peking University 2012--2013; Assistant Professor, Université de Montréal
- Kohinoor Dasgupta, PhD 2012, joint with Vijay Nair and Stilian Stoev;
Senior Biostatistician, Novartis India
- Cen Guo, PhD 2012, joint with Tailen Hsing; Data Scientist, Uber, CA
- Bopeng Li, MS 2012; in Statistics PhD program, University of Michigan
- Hyun-Chul Kim, Visiting Scholar 2010--2011; Research Professor, Yonsei University, Korea
Undergraduate honor thesis advisees
- 2017--2018: Jiahui Ji (UM Biostatistics), Zui Chen (Parsons School of Design)
- 2018--2019: Yingsi Jian, Jiayue Lu
- Editorial board
- Formative education in Hai Phong (Vietnam), undergraduate study at POSTECH (Korea)
- Master's degree from Arizona State University, apprenticing with Subbarao Kambhampati
- Ph.D, University of California, Berkeley, advised by Michael Jordan and Martin Wainwright
- Postdoctoral fellow at SAMSI and Duke University, mentored by Alan Gelfand and Jim Clark
On posterior contraction of parameters and interpretability in Bayesian mixture modeling.
A. Guha, N. Ho and X. Nguyen.
On functional aggregate queries with additive inequalities.
M. Abo Khamis, R. Curtin, B. Moseley, H. Ngo, X. Nguyen, D. Olteanu and M. Schleich.
Local inversion-free estimation of spatial Gaussian processes.
H. Keshavarz, X. Nguyen and C. Scott.
Robust estimation of mixing measures in finite mixture models.
N. Ho, X. Nguyen and Y. Ritov.
arXiv:1709.08094. To appear, Bernoulli.
Learning models over relational data using sparse tensors and functional dependencies.
M. Abo Khamis, H. Q. Ngo, X. Nguyen, D. Olteanu and M. Schleich.
Conic scan-and-cover algorithms for nonparametric topic modeling.
M. Yurochkin, A. Guha and X. Nguyen.
Advances in NIPS 30, 2017.
Multi-way interacting regression via factorization machines.
M. Yurochkin, X. Nguyen and N. Vasiloglou.
Advances in NIPS 30, 2017.
Multilevel clustering via Wasserstein means.
N. Ho, X. Nguyen, M. Yurochkin, H. H. Bui, V. Huynh and D. Phung.
Proceedings of the ICML, 2017.
Singularity structures and impacts on parameter estimation in
finite mixtures of distributions.
N. Ho and X. Nguyen. arXiv:1609.02655.
Borrowing strength in hierarchical Bayes: posterior concentration of the Dirichlet base measure.
X. Nguyen. Bernoulli, 22(3), 1535--1571, 2016.
Geometric Dirichlet means algorithm for topic inference.
M. Yurochkin and X. Nguyen.
Advances in NIPS 29, 2016.
Scalable nonparametric Bayesian multilevel clustering.
V. Huynh, D. Phung, S. Venkatesh, X. Nguyen, M. Hoffman and H. H. Bui.
Proceedings of UAI, 2016.
On the consistency of inversion-free parameter estimation for Gaussian random fields.
H. Keshavarz, C. Scott and X. Nguyen.
Journal of Multivariate Analysis, 150, 245--266, 2016.
Convergence rates of parameter estimation for some weakly identifiable finite mixtures.
N. Ho and X. Nguyen. Annals of Statistics, 44(6), 2726--2755, 2016.
On strong identifiability and convergence rates of parameter estimation in finite mixtures.
N. Ho and X. Nguyen. Electronic Journal of Statistics, 10(1), 271--307, 2016.
Optimal change point detection in Gaussian processes.
H. Keshavarz, C. Scott and X. Nguyen.
To appear, Journal of Statistical Planning and Inference.
Posterior contraction of the population polytope in finite admixture
X. Nguyen. Bernoulli, 21(1), 618--646, 2015.
Selected talk slides
Elements of data science
Summer School on Data Science,
Vietnam Institute for Advanced Study in Mathematics,
Hanoi and Ho Chi Minh, May 2017.
Multi-level clustering with contexts via hierarchical nonparametric Bayesian inference.
Biostatistics Seminar, University of Michigan, October 2016.
Singularity structures and parameter estimation in finite mixture models.
Workshop on Empirical Likelihood Methodology,
National University of Singapore, June 2016.
Topic modeling with more confidence: a theory and some algorithms.
Keynote talk, Pacific-Asia Knowledge Discovery and Data Mining Conference, Ho Chi Minh, May 2015.
Borrowing strength in hierarchical Bayes: convergence of the Dirichlet base measure.
9th Bayesian Nonparametrics Conference, Amsterdam, June 2013.
Convergence of latent mixing measures in finite and infinite mixture
models. Bayesian Nonparametrics Workshop at ICERM, Providence,
Clustering problems, mixture models and Bayesian nonparametrics.
VIASM Summer School, Hanoi, July 2012.
[Additional notes ]
Message-passing sequential detection of multiple change points
in networks. IEEE Symposium on Information Theory, Boston,
Inference of functional clusters from non-functional data .
Midwest Statistics Research Colloquium, Madison, March 2012.
Dirichlet labeling and hierarchical processes for clustering functional data
. IMS-China Conference, Xi'an, July 2011.
Decentralized decision making with spatially distributed data .
AI Seminar, University of Michigan, Oct 2009.
Surrogate loss functions, divergences and decentralized detection.
Thesis Talk, UC Berkeley, May 2007.
Anomaly and sequential detection with time series data .
Tutorial lectures given at Berkeley, 2006.