Advice-Efficient Prediction with Expert Advice

TL;DR

This paper introduces an algorithm for advice-efficient prediction with expert advice, reducing the number of expert advice queries per round while maintaining low regret, which is especially useful when querying all experts is costly.

Contribution

The paper proposes a novel algorithm that achieves sublinear regret in advice-efficient prediction with limited expert advice queries.

Findings

Achieves $O(\sqrt{rac{N}{M}T\ln N})$ regret bound.

Effectively reduces advice complexity while maintaining performance.

Applicable in scenarios with costly expert advice queries.

Abstract

Advice-efficient prediction with expert advice (in analogy to label-efficient prediction) is a variant of prediction with expert advice game, where on each round of the game we are allowed to ask for advice of a limited number $M$ out of $N$ experts. This setting is especially interesting when asking for advice of every expert on every round is expensive. We present an algorithm for advice-efficient prediction with expert advice that achieves $O(\sqrt{\frac{N}{M}T\ln N})$ regret on $T$ rounds of the game.