Publications

In Preparation:

• A New Perspective on Least-Norm Interpolation Under Gaussian Covariates

Gil Kur*, Zong Shang, Paul Simanjuntak, Guillaume Lecue, and Reese Pathak*

Preprints:

• Early-Stopping in Linear Least Squares over Convex Bodies 
  Tobias Wegel*, Gil Kur, and Patrick Rebeschini

• Optimality of Maximum Likelihood for Log-Concave Density Estimation and Convex Regression 
  Gil Kur*, Yuval Dagan, and Alexander Rakhlin  
  
  

Publications:

Statistics and Machine Learning:

•  Projection Pursuit in High Dimensions
  Boaz Nadler, Gil Kur*, and Peter J. Bickel (alphabetical)
  Proceedings of the National Academy of Sciences

• Space Lower Bounds for Linear Prediction
  Yuval Dagan*, Gil Kur, and Ohad Shamir 
  Conference on Learning Theory 2019

• Convex Regression in Multidimensions: Suboptimality of Least Squares Estimators
  Gil Kur*, Fuchang Gao, Adityanand Guntuboyina, and Bodhi Sen
  Annals of Statistics 

• On Suboptimality of Least Squares with Application to Estimation of Convex Bodies 
  Gil Kur*, Adityanand Guntuboyina, and Alexander Rakhlin
  Conference on Learning Theory 2020

• On the Minimal Error of Empirical Risk Minimization
  Gil Kur* and Alexander Rakhlin 
  Conference on Learning Theory 2021
  Video

• A Bounded-Noise Mechanism for Differential Privacy
  Yuval Dagan* and Gil Kur
  Conference on Learning Theory 2022

• An Efficient Minimax Optimal Estimator for Multivariate Convex Regression
  Gil Kur* and Eli Putterman
  Conference on Learning Theory 2022
  Slides (on estimating a norm of a convex function via  a  smart averaging)

• Tyler's and Maronna's M-estimators: Non-Asymptotic Concentration Results
  Elad Romanov*, Gil Kur, and Boaz Nadler 
  Journal of Multivariate Analysis 

•  On the Variance, Admissibility, and Stability of Empirical Risk Minimization
  Gil Kur*, Eli Putterman, and Alexander Rakhlin 
  Neurips 2023 (spotlight)
  Video (presented by Eli Putterman)

• Minimum Norm Interpolation Meets The Local Theory of Banach Spaces 
  Gil Kur* and Pedro Abdalla, Pierre Bizeul, and Fanny Yang 
  International Conference on Machine Learning 2024
  Video

• Adaptive Convergence Rates for Log-Concave Maximum Likelihood
  Gil Kur* and Adityanand Guntuboyina
  AISTATS 2024

• Debiased LASSO under Poisson-Gauss Model
  Pedro Abdalla* and Gil Kur (alphabetical)
  Information and Inference: A Journal of the IMA

• Revisiting Knowledge Distillation: surprising gains in data efficiency 
  Giulia Lanzillotta*, Felix Sarnthein, Gil Kur, Bobby He, and Thomas Hofmann
  NeurIPS 2024 Workshop on Scientific Methods for Understanding Deep Learning

• Convergence rates for estimating multivariate scale mixtures of uniform densities
  Arlene KH Kim*, Gil Kur, Adityanand Guntuboyina
  Electronic Journal of Statistics

•  Specialization after Generalization: Towards Understanding Test-Time Training in Foundation Models
  Jonas Hübotter*, Patrick Wolf*, Alexander Shevchenko*, Dennis Jüni , Andreas Krause, and Gil Kur
  Oral at NeurIPS 2025 CCFM Workshop, submitted to ICLR 2026
  
  

Convex Geometry:

• Approximation of the Euclidean ball by polytopes with a restricted number of facets
  Gil Kur
  Studia Mathematica 

•  A Concentration Inequality for Random Polytopes, Dirichlet-Voronoi Tiling Numbers and the Geometric Balls and Bins Problem
  Steven Hoehner and Gil Kur* (alphabetical)
  Discrete & Computational Geometry 

• Intrinsic and dual volume deviations of convex bodies and polytopes 
  Florian Besau, Steven Hoehner*, and Gil Kur (alphabetical)
  International Mathematical Research Notes
  Slides (by Steven Hoehner)