unlike those results, this bound requires no additional assumptions on the domain or on the magni- tude/direction of the accumulated gradients

hence ln(1/wσs 1 )≤2 ln(s+ 1)≤2 ln(log( p 1 + 6PT/R+ 1)) I Additional Details on Laziness Advantages I · 2010

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

1 Pith paper citing it

browse 1 citing papers

representative citing papers

cs.LG · 2026-01-22 · unverdicted · novelty 6.0

k-lazyGD achieves optimal dynamic regret O(sqrt((P_T+1)T)) in SOCO for laziness k up to Theta(sqrt(T/P_T)).

Showing 1 of 1 citing paper.

Partially Lazy Gradient Descent for Smoothed Online Learning cs.LG · 2026-01-22 · unverdicted · none · ref 13
k-lazyGD achieves optimal dynamic regret O(sqrt((P_T+1)T)) in SOCO for laziness k up to Theta(sqrt(T/P_T)).