Year: 2026
Type: preprint
Abstract: We prove a tight $Θ(n/ε)$ lower bound on the number of samples required for testing halfspaces over $\mathbb{R}^n$, in the distribution-free sample-based model where the underlying probability distribution is unknown to the algorithm, and the algorithm only receives random samples (i.e., it cannot make queries). This shows that testing is no more efficient than learning for halfspaces. We also show a matching upper bound for one-sided testers, improving on the standard (two-sided) testing-by-learning reduction, establishing that two-sided halfspace testers in this model have no advantage over one-sided testers.
Source: arXiv (Cornell University)
Institutions: Columbia University, University of British Columbia
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7202345381
Cites:
Cited by:
Related to: