2018 George Nicholson Paper Award

News - 11 December 2018

2018 George Nicholson Paper Award

The George Nicholson Student Paper Competition, arguably the most prestigious student award in the operations research community, is held each year since 1975 to honor outstanding student papers in the field of operations research and the management sciences. Viet Anh, a visiting PhD student at TU Delft and co-supervised by Peyman Mohajerin Esfahani, is the first winner from a European university. He received the 2018 Nicholson Award for the paper “Distributionally Robust Inverse Covariance Estimation: The Wasserstein Shrinkage Estimator”.

The precision matrix is a key ingredient for a number of important problems including the optimal feedback control in the LQR problem, the optimal classification rule in linear discriminant analysis, the optimal investment portfolio in Markowitz’ celebrated mean-variance model, and the optimal array vector of the beamforming problem in signal processing. Moreover, the optimal fingerprint method used to detect a multivariate climate change signal blurred by weather noise requires knowledge of the climate vector’s precision matrix. In this paper, it is shown that the proposed estimator has many desirable theoretical properties and displays a similar performance as the graphical lasso approach (which requires solving a large semidefinite program) at the computational cost of a naive linear shrinkage estimator (which requires merely a spectral decomposition).

A follow-up of this paper offers a game-theoretic setting to propose a novel robust Kalman filter for state-estimation in dynamical systems. This study has also been featured as a spotlight presentation (acceptance rate: 3.5%) at the 2018 Conference on Neural Information Processing Systems (NIPS), the leading conference in machine learning.

Related link and papers:

-        Award link: https://www.informs.org/Recognizing-Excellence/Award-Recipients/Viet-Anh-Nguyen

-        George Nicholson paper: “Distributionally Robust Inverse Covariance Estimation: The Wasserstein Shrinkage Estimator”, Viet Anh Nguyen, Daniel Kuhn, and Peyman Mohajerin Esfahani, May 2018, [arXiv:1805.07194]

-        NIPS spotlight presentation paper: “Wasserstein Distributionally Robust Kalman Filtering”, Soroosh Shafieezadeh-Abadeh, Viet Anh Nguyen, Daniel Kuhn, and Peyman Mohajerin Esfahani, Neural Information Processing Systems (NIPS), December 2018, [arXiv: 1809.08830]