Then, we have ∥z1 −z 2∥2 A1 +∥z 1 −z 2∥2 A2 ≤ ⟨w1 −w 2, z2 −z 1⟩+ (ψ 1(z2)−ψ 2(z2))−(ψ 1(z1)−ψ 2(z1))

Suppose thatz 1 ∈arg min x∈X ⟨w1, x⟩+ψ1(x) andz 2 ∈arg min x∈X ⟨w2, x⟩+ψ2(x) · 2007

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

browse 1 citing papers

representative citing papers

Exploiting Curvature in Online Convex Optimization with Delayed Feedback

cs.LG · 2025-06-09 · accept · novelty 7.0

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.

citing papers explorer

Showing 1 of 1 citing paper.

Exploiting Curvature in Online Convex Optimization with Delayed Feedback cs.LG · 2025-06-09 · accept · none · ref 2
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.

Then, we have ∥z1 −z 2∥2 A1 +∥z 1 −z 2∥2 A2 ≤ ⟨w1 −w 2, z2 −z 1⟩+ (ψ 1(z2)−ψ 2(z2))−(ψ 1(z1)−ψ 2(z1))

fields

years

verdicts

representative citing papers

citing papers explorer