An online potential-based algorithm achieves O(sqrt n) terminal discrepancy with exponentially high probability for independent sub-Gaussian inputs, and a sparsity-aware variant achieves O(sqrt k).
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DS 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Online Discrepancy Minimization for Sub-Gaussian Inputs via Regularization and Restriction
An online potential-based algorithm achieves O(sqrt n) terminal discrepancy with exponentially high probability for independent sub-Gaussian inputs, and a sparsity-aware variant achieves O(sqrt k).