The Pareto Frontier of Randomized Learning-Augmented Online Bidding

2026-05-07

AI Summary

This work addresses the trade-off between leveraging predictions and ensuring robustness-consistency guarantees in online bidding. The authors propose a randomized, learning-augmented bidding strategy that integrates oracle predictions through a novel abstraction termed the “bidding function,” unifying the modeling and analysis of randomized bidding policies. Their approach closes the theoretical gap between upper and lower bounds for competitive ratios when \( R \geq 2.885 \), establishing for the first time a tight Pareto-optimal frontier between robustness and consistency. Empirical validation on the incremental k-median clustering task demonstrates the practical efficacy of the proposed method.

Abstract

Online bidding is a classical problem in online decision-making, with applications in resource allocation, hierarchical clustering, and the analysis of approximation algorithms. We study its randomized learning-augmented variant, where an online algorithm generates a sequence of random bids while leveraging predictions from an oracle. We provide analytical upper and lower bounds on the optimal consistency $C$ as a function of the robustness $R$, which match when $R \geq 2.885$, effectively closing the gap left by previous work. The key technical ingredient is the notion of a bidding function, a novel abstraction that provides a unified framework for the design and analysis of randomized bidding strategies. We complement our theoretical results with an experimental application of randomized bidding to the incremental median problem, demonstrating the applicability of our algorithm in practical clustering settings.

Problem

• online bidding
• randomized algorithms
• learning-augmented algorithms
• Pareto frontier
• consistency-robustness tradeoff

Innovation

• randomized bidding
• learning-augmented algorithms
• Pareto frontier
• bidding function
• online decision-making