A Wasserstein distributionally robust version of the performance estimation problem is minimized to learn robust hyperparameters for first-order methods, unifying data-driven L2O and worst-case PEP design.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Distributionally-Robust Learning to Optimize
A Wasserstein distributionally robust version of the performance estimation problem is minimized to learn robust hyperparameters for first-order methods, unifying data-driven L2O and worst-case PEP design.