Assistant Professor
Education
- Ph.D. Applied Mathematics (2018), Université Grenoble Alpes
- M.Sc. Applied Mathematics and Physics (2014), Moscow Institute of Physics and Technology
- B.Sc. Applied Mathematics and Physics (2012), Moscow Institute of Physics and Technology
Expertise
- Numerical Optimization
- Mathematical Statistics
- Data Science and Machine Learning
- Information theory
About
Dmitrii Ostrovskii is a tenure-track Assistant Professor with the joint appointment in the School of Mathematics and H. Milton Stewart School of Industrial and Systems Engineering. Prior to joining Georgia Tech, he completed postdocs at the University of Southern California (2019-2023) and INRIA National Research Institute in Paris (2018-2019).
Research
Dmitrii Ostrovskii is an Assistant Professor with a joint appointment in the Schools of Mathematics and Industrial and Systems Engineering at Georgia Tech, since Fall 2023. He earned his Bachelor and Master degrees from the Moscow Institute of Physics and Technology and PhD from Grenoble-Alpes University; he held postdoctoral appointments at the University of Southern California and at the National Institute of Research in Digital Science and Technology (INRIA) in Paris, France.
Dr. Ostrovskii's research is in mathematical statistics, optimization theory, and theoretical foundations of data science. Across these fields, he made essential contributions in several areas of statistical theory and optimization, such as online optimization, minimax optimization, statistically and computationally efficient estimation, and inference under structural constraints.. Dr. Ostrovskii's research is focused on providing near-optimal algorithms and inference methods for the problems that combine theoretical challenges with sound practical motivation -- in particular, those arising in data-intensive applications, where one aims to reveal the structure hidden in extremely large and high-dimensional datasets, and do so in a computationally tractable way. More recently, his focus has shifted towards quantum information science (including tomography and state estimation), trigonometric interpolation, problems at the intersection of approximation theory and algebraic combinatorics, as well as optimization problems arising in electrical power grids.
Teaching
Since he joined Georgia Tech, Dr. Ostrovskii has taught advanced-level classes, often with mixed enrollment from across the campus (Math, ISyE, and ECE):
- ISyE 8803: Advanced Tools and Special Topics in Mathematical Data Science (x2)
- Math 7252: High-Dimensional Statistics
- Math 6252: Statistical Estimation
The first of these has been developed by Dr. Ostrovskii from scratch. He has also taught undergraduate classes in linear algebra and differential equations (Math 1552 & 2552).
Representative Publications
- D. Ostrovskii, P. Shcherbakov. Amplitude Maximization in Stable Systems, Schur Positivity, and Some Conjectures on Polynomial Interpolation. arXiv:2508.13554
- D. Ostrovskii. Near-Optimal and Tractable Estimation under Shift-Invariance. arXiv:2411.03383
- S. Talkington, D. Ostrovskii, D. Molzahn. Efficient Network Reconfiguration by Randomized Switching. arXiv:2510.24458
- S. Talkington, C. Khanpour, R. Gupta, S. Dorado-Rojas, D. Turizo, H. Park, D. Ostrovskii, D. Molzahn. Admittance Matrix Concentration Inequailties for Uncertain Power Networks arXiv:2510.17798
Nonconvex-Nonconcave Min-Max Optimization with a Small Maximization Domain
D. Ostrovskii, B. Barazandeh, M. Razaviyayn arXiv:2110.03950, 2021
Near-Optimal Procedures for Model Discrimination with Non-Disclosure Properties
D. Ostrovskii, M. Ndaoud, A. Javanmard, M. Razaviyayn arXiv:2012.02901, 2020
Efficient Primal-Dual Algorithms for Large-Scale Multiclass Classification
D. Babichev, D. Ostrovskii, F. Bach arXiv:1902.03755, 2019
Book ChapterAdaptive Denoising of Signals with Shift-Invariant Structure
D. Ostrovskii, Z. Harchaoui, A. Judistky, A. Nemirovski arXiv:1806.04028
Foundations of Modern Statistics: V. Spokoiny’s 60th Anniversary Festscrift
Journal PapersEfficient and Near-Optimal Online Portfolio Selection
R. Jézéquel, D. Ostrovskii, P. Gaillard arXiv:2209.13932
Mathematics of Operations Research, 2025 (to appear)
Efficient Search of First-Order Nash Equilibria in Nonconvex-Concave Smooth
Min-Max Problems
D. Ostrovskii, A. Lowy, M. Razaviyayn arXiv:2002.07919
SIAM Journal on Optimization, 31:4, pp. 2508-2538, 2021
Finite-Sample Analysis of M-Estimators Using Self-Concordance
D. Ostrovskii, F. Bach arXiv:1810.06838
Electronic Journal of Statistics, 15:1, pp. 326-391, 2021
Concentration Inequalities for the Exponential Weighting Method
Affine Invariant Covariance Estimation for Heavy-Tailed Distributions
D. Ostrovskii, A. Rudi arXiv:1902.03086
COLT 2019
Beyond Least-Squares: Fast Rates for Regularized Empirical Risk Minimization through
Self-Concordance
U. Marteau-Ferey, D. Ostrovskii, A. Rudi, F. Bach arXiv:1902.03046
COLT 2019