Reese Pathak

NSF Postdoc • Berkeley Statistics • Cornell ORIE

UC Berkeley, Department of Statistics
Cornell University, School of Operations Research and Information Engineering (ORIE)

E-mail: pathakr (at) berkeley (dot) edu

I am an NSF Mathematical Sciences Postdoctoral Research Fellow (MSPRF) at UC Berkeley, with sponsoring scientist Nikita Zhivotovskiy, and a Visiting Assistant Professor at Cornell University. I received my PhD in Electrical Engineering and Computer Sciences (EECS) in 2025 from UC Berkeley, advised by Michael I. Jordan and Martin J. Wainwright.

In January 2027, I will join Cornell ORIE as a tenure-track assistant professor.

Interests: high-dimensional phenomena in statistics, probability, and geometry.

Publications and preprints

Beyond modern asymptotics for log-likelihood ratios in logistic regression (with Hugo Chardon and Nikita Zhivotovskiy) [ arXiv ]
Optimal mean width and metric entropy estimates for convex bodies (with Grigoris Paouris) [ arXiv ]
Minimum norm interpolation via the local theory of Banach spaces: the role of Gaussianity (with Gil Kur) [ arXiv ]
On the metric projection onto a convex set: reverse Hölder inequalities and upper bounds [ arXiv ]
A remark on the majorizing measures theorem for general processes (with Nikita Zhivotovskiy) [ arXiv ]
Gaussian width of convex sets via integral decompositions, projections, and the distribution of intrinsic volumes (with Nikita Zhivotovskiy) [ arXiv ]
A new perspective on minimum-norm interpolation under Gaussian covariates (with Gil Kur, Zong Shang, Paul Simanjuntak, and Guillaume Lecué) [ AISTATS ]
Revisiting mean estimation over ℓ_p balls: is the MLE optimal? (with Liviu Aolaritei, Michael I. Jordan, and Annie Ulichney) [ arXiv ]
On the design-dependent suboptimality of the Lasso (with Cong Ma) [ arXiv ]
Transformers can optimally learn regression mixture models (with Rajat Sen, Weihao Kong, and Abhimanyu Das) [ arXiv ] [ ICLR ]
Noisy recovery from random linear observations: sharp minimax rates under elliptical constraints (with Martin J. Wainwright and Lin Xiao) [ arXiv ] [ Annals of Statistics ]
Optimally tackling covariate shift in RKHS-based nonparametric regression (with Cong Ma and Martin J. Wainwright) [ arXiv ] [ Annals of Statistics ]
A new similarity measure for covariate shift with applications to nonparametric regression (with Cong Ma and Martin J. Wainwright) [ arXiv ] [ ICML (long oral) | slides | poster ]
Weighted matrix completion from non-random, non-uniform sampling patterns (with Simon Foucart, Deanna Needell, Yaniv Plan, and Mary Wootters) [ arXiv ] [ IEEE Transactions on Information Theory ]
FedSplit: an algorithmic framework for fast federated optimization (with Martin J. Wainwright) [ arXiv ] [ NeurIPS ]