Delayed-feedback FTRL, Online Newton Step, and a clipped Vovk-Azoury-Warmuth forecaster achieve regret of order min{logarithmic in maximum backlog, square root of total delay} for strongly convex, exp-concave, and online linear regression losses.
(2024) only manage to show that this algorithm achieves regret O dmax(G2+D) λ lnT+ dmaxG λ2 under Theorems 2.3 and 2.4
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
Exploiting Curvature in Online Convex Optimization with Delayed Feedback
Delayed-feedback FTRL, Online Newton Step, and a clipped Vovk-Azoury-Warmuth forecaster achieve regret of order min{logarithmic in maximum backlog, square root of total delay} for strongly convex, exp-concave, and online linear regression losses.