Adversarial Delays in Online Strongly-Convex Optimization

by   Daniel Khashabi, et al.

We consider the problem of strongly-convex online optimization in presence of adversarial delays; in a T-iteration online game, the feedback of the player's query at time t is arbitrarily delayed by an adversary for d_t rounds and delivered before the game ends, at iteration t+d_t-1. Specifically for online-gradient-descent algorithm we show it has a simple regret bound of ∑_t=1^T (1+ d_t/t). This gives a clear and simple bound without resorting any distributional and limiting assumptions on the delays. We further show how this result encompasses and generalizes several of the existing known results in the literature. Specifically it matches the celebrated logarithmic regret T when there are no delays (i.e. d_t = 1) and regret bound of τ T for constant delays d_t = τ.


Online Strongly Convex Optimization with Unknown Delays

We investigate the problem of online convex optimization with unknown de...

No Discounted-Regret Learning in Adversarial Bandits with Delays

Consider a player that in each round t out of T rounds chooses an action...

Online Frank-Wolfe with Unknown Delays

The online Frank-Wolfe (OFW) method has gained much popularity for onlin...

Temporal Connectivity: Coping with Foreseen and Unforeseen Delays

Consider planning a trip in a train network. In contrast to, say, a road...

Robust Online Convex Optimization in the Presence of Outliers

We consider online convex optimization when a number k of data points ar...

Dynamic Online Gradient Descent with Improved Query Complexity: A Theoretical Revisit

We provide a new theoretical analysis framework to investigate online gr...

Learning Under Delayed Feedback: Implicitly Adapting to Gradient Delays

We consider stochastic convex optimization problems, where several machi...