Publications and Preprints
An up-to-date list of my publications can be found on arXiv and on my Google scholar page.
-
L. Chizat and T. Vaškevičius. Computational guarantees for doubly entropic Wasserstein barycenters via damped Sinkhorn Iterations. Conference on Neural Information Processing Systems (NeurIPS), vol. 36, pp. 12368-12388, 2023, arXiv 2307.13370, link to proceedings, link to code.
-
J. Mourtada, T. Vaškevičius, N. Zhivotovskiy. Local risk bounds for statistical aggregation. Proceedings of Thirty Sixth Conference on Learning Theory (COLT), PMLR 195:5697-5698, 2023, arXiv 2306.17151, link to proceedings.
-
V. Kanade, P. Rebeschini, T. Vaškevičius. Exponential tail local Rademacher complexity risk bounds without the Bernstein condition. Preprint, 2022, arXiv 2202.11461.
-
J. Mourtada, T. Vaškevičius, N. Zhivotovskiy, Distribution-free robust linear regression. Mathematical Statistics and Learning, vol. 4, no. 3, pp. 253-292, 2022, arXiv 2102.12919, link to journal.
-
T. Vaškevičius, N. Zhivotovskiy. Suboptimality of constrained least squares and improvements via non-linear predictors. Bernoulli, vol. 29, no. 1, pp. 473-495, 2023, arXiv 2009.09304, link to journal.
-
V. Kanade, P. Rebeschini, T. Vaškevičius. The statistical complexity of early-stopped mirror descent. Information and Inference: A Journal of the IMA, vol. 12, pp. 3010-3041, 2023, arXiv 2002.00189, link to journal, link to code.
Earlier version appeared at NeurIPS 2020 (spotlight presentation).
-
V. Kanade, P. Rebeschini, T. Vaškevičius. Implicit regularization for optimal sparse recovery. Conference on Neural Information Processing Systems (NeurIPS), vol. 32, pp. 2972-2983, 2019, arXiv 1909.05122, link to proceedings, link to code.