Tractability from overparametrization: The example of the negative perceptron
Abstract
In the negative perceptron problem we are given data points , where is a -dimensional vector and is a binary label. The data are not linearly separable and hence we content ourselves to find a linear classifier with the largest possible \emph{negative} margin. In other words, we want to find a unit norm vector that maximizes . This is a non-convex optimization problem (it is equivalent to finding a maximum norm vector in a polytope), and we study its typical properties under two random models for the data. We consider the proportional asymptotics in which with , and prove upper and lower bounds on the maximum margin or -- equivalently -- on its inverse function . In other words, is the overparametrization threshold: for a classifier achieving vanishing training error exists with high probability, while for it does not. Our bounds on match to the leading order as . We then analyze a linear programming algorithm to find a solution, and characterize the corresponding threshold . We observe a gap between the interpolation threshold and the linear programming threshold , raising the question of the behavior of other algorithms.
Cite
@article{arxiv.2110.15824,
title = {Tractability from overparametrization: The example of the negative perceptron},
author = {Andrea Montanari and Yiqiao Zhong and Kangjie Zhou},
journal= {arXiv preprint arXiv:2110.15824},
year = {2025}
}
Comments
107 pages; 7 pdf figures