Trading-Off Payments and Accuracy in Online Classification with Paid Stochastic Experts

TL;DR

This paper presents an online learning algorithm for classification with paid stochastic experts, balancing payment costs and prediction accuracy, and achieves near-optimal regret bounds by combining Lipschitz bandits with surrogate losses.

Contribution

It introduces a novel algorithm that optimally balances payments and accuracy in online classification with stochastic experts, improving regret bounds over standard Lipschitz bandit approaches.

Findings

The algorithm achieves a regret of O(K^2(log T)√T).

It outperforms standard Lipschitz bandit bounds in experiments.

Empirical evaluation on synthetic data confirms effectiveness.

Abstract

We investigate online classification with paid stochastic experts. Here, before making their prediction, each expert must be paid. The amount that we pay each expert directly influences the accuracy of their prediction through some unknown Lipschitz "productivity" function. In each round, the learner must decide how much to pay each expert and then make a prediction. They incur a cost equal to a weighted sum of the prediction error and upfront payments for all experts. We introduce an online learning algorithm whose total cost after $T$ rounds exceeds that of a predictor which knows the productivity of all experts in advance by at most $\mathcal{O}(K^2(\log T)\sqrt{T})$ where $K$ is the number of experts. In order to achieve this result, we combine Lipschitz bandits and online classification with surrogate losses. These tools allow us to improve upon the bound of order $T^{2/3}$ one would obtain in the standard Lipschitz bandit setting. Our algorithm is empirically evaluated on synthetic data