Statistics Theory

Arash A. Amini, Bryon Aragam, Qing Zhou

This work addresses the problem of learning multivariate dependencies in nonparametric and high-dimensional settings, offering a unified approach that works without faithfulness and avoids the curse of dimensionality.

Matthew S. Zhang, Jason M. Altschuler, Sinho Chewi

This resolves the computational bottleneck of finding warm starts for HMC, which is crucial for practitioners in statistics, engineering, and sciences who rely on HMC for high-dimensional sampling, though it is incremental as it builds on prior theoretical work.

Steve Hanneke, Anay Mehrotra, Grigoris Velegkas et al.

Provides a theoretical characterization and first algorithm for a classic but understudied learning model, clarifying the role of membership queries.

Tor Lattimore

This provides theoretical insights into the performance of policy gradient in bandit settings, which is incremental as it builds on existing methods with new analysis.

Boris Hanin, Tianze Jiang

This work provides theoretical insights into the behavior of deep non-linear MLPs for researchers and practitioners interested in understanding the benefits of network depth in Bayesian inference, particularly in large-scale regimes.

Charlie Cowen-Breen, Alekh Agarwal, Stephen Bates et al.

Provides a general framework for resource-constrained statistical estimation, improving efficiency for practitioners using multiple proxies.

Xifan Yu, Ilias Zadik

This work provides rigorous algorithmic foundations for the physics belief that first-order phase transitions impose fundamental limits on efficient algorithms, addressing a long-standing problem in theoretical computer science and statistical estimation.

Nikhil Bansal, Matthias C. Caro, Gaurav Mahajan

It establishes a fundamental equivalence between cloning and learning for an important class of quantum states, providing a fine-grained perspective on No-Cloning theorems with implications for quantum learning theory and cryptography.

Reese Pathak, Nikita Zhivotovskiy · eth-zurich

This extends a fundamental inequality from Gaussian processes to a much larger family, providing a unified theoretical foundation for empirical process theory.

Anay Mehrotra, Phuc Tran, Van H. Vu et al.

For researchers in causal inference and matrix completion, this work offers the first sharp row-wise ℓ2 perturbation bound, enabling per-unit treatment effect estimation where previous methods only gave average bounds.

Jingda Wu, Changxiao Cai

Provides the first rigorous theoretical justification for diffusion models' ability to adapt to low-dimensional structure and multi-modality, addressing a key gap in understanding their empirical success.

Kelvin Kan, Xingjian Li, Benjamin J. Zhang et al.

This work provides the first convergence theory for discrete diffusion models that scales to large vocabularies (e.g., hundreds of thousands of tokens) by removing the state-space-size dependence that made prior bounds vacuous for modern language tasks.

Zhiyang Xun, Eric Price

Provides fundamental limits for diffusion sampling acceleration, relevant to researchers designing faster sampling algorithms.

Liu Zhang, Amit Singer

For practitioners using moment-based estimation in the presence of outliers, this work provides a theoretically grounded robust method with finite-sample guarantees.

Nathanael Tepakbong, Hanyu Hu, Chengyu Liu et al.

For researchers using PINNs to solve PDEs, this work provides a theoretically grounded method to mitigate training difficulties caused by ill-conditioned loss landscapes, with practical improvements on several benchmark problems.

Jason M. Altschuler, Sinho Chewi, Matthew S. Zhang

This work addresses a major open problem in computational statistics and machine learning by enabling faster sampling algorithms for log-concave distributions, which is incremental but crucial for applications like Bayesian inference.

Santosh S. Vempala, Andre Wibisono

This provides a substantial generalization of sampling methods for convex and star-shaped bodies, addressing a fundamental challenge in computational geometry and statistics.

Danru Xu, Sébastien Lachapelle, Sara Magliacane

For researchers in causal representation learning, this work extends identifiability guarantees to degenerate Gaussian mixtures, addressing a challenging setting where standard density-based methods fail.

Krishnakumar Balasubramanian

This work provides theoretical convergence guarantees for a new class of generative models, addressing conservatism issues in drifting methods, but the rates are limited by curse of dimensionality.

Tristan Luca Saidi, Gonzalo Mena, Larry Wasserman et al.

This addresses the challenge of modeling distributional dynamics in scientific systems where classical vector space methods fail, offering a novel approach for researchers in fields like biology and economics.