Disordered systems insights on computational hardness

D Gamarnik, C Moore… - Journal of Statistical …, 2022 - iopscience.iop.org
In this review article we discuss connections between the physics of disordered systems,
phase transitions in inference problems, and computational hardness. We introduce two …

A precise high-dimensional asymptotic theory for boosting and minimum--norm interpolated classifiers

T Liang, P Sur - The Annals of Statistics, 2022 - projecteuclid.org
A precise high-dimensional asymptotic theory for boosting and minimum-l1-norm
interpolated classifiers Page 1 The Annals of Statistics 2022, Vol. 50, No. 3, 1669–1695 …

Computational barriers to estimation from low-degree polynomials

T Schramm, AS Wein - The Annals of Statistics, 2022 - projecteuclid.org
Computational barriers to estimation from low-degree polynomials Page 1 The Annals of
Statistics 2022, Vol. 50, No. 3, 1833–1858 https://doi.org/10.1214/22-AOS2179 © Institute of …

Lattice-based methods surpass sum-of-squares in clustering

I Zadik, MJ Song, AS Wein… - Conference on Learning …, 2022 - proceedings.mlr.press
Clustering is a fundamental primitive in unsupervised learning which gives rise to a rich
class of computationally-challenging inference tasks. In this work, we focus on the canonical …

Sum-of-squares lower bounds for densest k-subgraph

C Jones, A Potechin, G Rajendran, J Xu - Proceedings of the 55th …, 2023 - dl.acm.org
Given a graph and an integer k, Densest k-Subgraph is the algorithmic task of finding the
subgraph on k vertices with the maximum number of edges. This is a fundamental problem …

Sum-of-squares lower bounds for sparse independent set

C Jones, A Potechin, G Rajendran… - 2021 IEEE 62nd …, 2022 - ieeexplore.ieee.org
The Sum-of-Squares (SoS) hierarchy of semidefinite programs is a powerful algorithmic
paradigm which captures state-of-the-art algorithmic guarantees for a wide array of …

Sub-exponential time Sum-of-Squares lower bounds for Principal Components Analysis

A Potechin, G Rajendran - Advances in Neural Information …, 2022 - proceedings.neurips.cc
Abstract Principal Components Analysis (PCA) is a dimension-reduction technique widely
used in machine learning and statistics. However, due to the dependence of the principal …

Non-gaussian component analysis via lattice basis reduction

I Diakonikolas, D Kane - Conference on Learning Theory, 2022 - proceedings.mlr.press
Abstract Non-Gaussian Component Analysis (NGCA) is the following distribution learning
problem: Given iid samples from a distribution on $\R^ d $ that is non-gaussian in a hidden …

Semidefinite programs simulate approximate message passing robustly

M Ivkov, T Schramm - Proceedings of the 56th Annual ACM Symposium …, 2024 - dl.acm.org
Approximate message passing (AMP) is a family of iterative algorithms that generalize
matrix power iteration. AMP algorithms are known to optimally solve many average-case …

Potential Hessian Ascent: The Sherrington-Kirkpatrick Model

D Jekel, JS Sandhu, J Shi - Proceedings of the 2025 Annual ACM-SIAM …, 2025 - SIAM
We present the first iterative spectral algorithm to find near-optimal solutions for a random
quadratic objective over the discrete hypercube, resolving a conjecture of Subag [Sub21] …