Near-Optimal Consistency-Robustness Trade-Offs for Learning-Augmented Online Knapsack Problems

TL;DR

This paper presents a family of learning-augmented algorithms for online knapsack problems that balance consistency and robustness near optimality, using practical predictions and a novel conversion method.

Contribution

It introduces a simple combination of trusted learning-augmented and worst-case algorithms achieving near Pareto-optimal trade-offs, along with a new fractional-to-integral conversion technique.

Findings

Achieves near Pareto-optimal consistency-robustness trade-offs

Uses practical predictions like single values or intervals for item values

Introduces a novel fractional-to-integral conversion procedure

Abstract

This paper introduces a family of learning-augmented algorithms for online knapsack problems that achieve near Pareto-optimal consistency-robustness trade-offs through a simple combination of trusted learning-augmented and worst-case algorithms. Our approach relies on succinct, practical predictions -- single values or intervals estimating the minimum value of any item in an offline solution. Additionally, we propose a novel fractional-to-integral conversion procedure, offering new insights for online algorithm design.