Student Seminar Series – Spring 2018

Schedule for Spring 2018


Seminars are on Wednesdays
Time: 12:00pm – 1:00pm
Location: Room 1025, 1255 Amsterdam Avenue
Contacts: Jonathan Auerbach

Information for speakers: For information about schedule, direction, equipment, reimbursement and hotel, please click here.


Richard Davis (Columbia)


Dong Xia (Columbia)

“Computationally Efficient Tensor Completion with Statistical Optimality.”

We develop methods for estimating a low rank tensor from noisy observations on a subset of its entries to achieve both statistical and computational efficiencies. There have been a lot of recent interests in this problem of noisy tensor completion. Much of the attention has been focused on the fundamental computational challenges often associated with problems involving higher order tensors, yet very little is known about their statistical performance. To fill in this void, in this article, we characterize the fundamental statistical limits of noisy tensor completion by establishing minimax optimal rates of convergence for estimating a k-th order low rank tensor which suggest significant room for improvement over the existing approaches. Furthermore, we propose a polynomial-time computable estimating procedure based upon power iteration and a second-order spectral initialization that achieves the optimal rates of convergence. Our method is fairly easy to implement and numerical experiments are presented to further demonstrate the practical merits of our estimator.


Tian Zheng (Columbia) “Creating a website with R blogdown”


Room 1025 SSW

Lutz Duembgen (University of Bern)

Title: Geodesic Convexity and Regularized Scatter Estimation

Abstract: In the first part of this talk we provide a brief introduction to a particular Riemannian geometry on the space of
symmetric, positive definite matrices. Then we introduce the notions of geodesic convexity and geodesic coercivity of functions on this
metric space. In the second part it is shown that the target functions underlying standard M-estimators of multivariate scatter are
geodesically convex and, under mild regularity conditions, even strictly geodesically convex and geodesically coercive. In high
dimensional settings, however, the latter conditions are necessarily violated which necessitates some sort of regularization. We present
some suitable geodesically convex penalty functions. The resulting regularized estimators with tuning parameter chosen by some cross
validation scheme are illustrated in a small simulation experiment.

(This is joint work with David Tyler, Heike Schuhmacher, Klaus Nordhausen, Markus Pauly and Thomas Schweizer.)

2/14/18  Wenda Zhou “Introduction to High Performance Computing”
2/21/18 Mitzi Morris “Data Structures and Algorithms for Efficient Statistical Computation”
2/28/18  Jonathan Auerbach “Publication Quality Plots with ggplot”

Eliot Gordon “University department housing, navigating the transfer process”

3/21/18  Phyllis Wan “Latex”
3/28/18  Gonzalo Mena (Columbia)
4/4/18   Florian Stebegg (Columbia) “Statistical Consulting”
4/11/18  Wenda Zhou (Columbia) “Tensor Flow”

Aaron Plasek (Columbia)

 4/25/18   David Hirshberg (Columbia) “Double Robustness” 
5/2/18 Charles Margossian (Columbia) “Automatic Differentiation: High Performance Implementation for Modern Statistical Problems in Academia and Industry with Applications to the Scientific, the Engineering, and the non-Scientific and non-Engineering fields: A Quantitative Approach with Implications for Dan Brownian Motion “