Improving Online Algorithms via ML Predictions

TL;DR

This paper explores how machine-learned predictions can enhance online algorithms for problems like ski rental and job scheduling, creating algorithms that adapt to prediction quality without significant performance loss.

Contribution

It introduces new online algorithms that leverage ML predictions, improving decision-making in classical problems while maintaining robustness against prediction errors.

Findings

Algorithms improve with better predictions

Performance degrades gracefully with poor predictions

Applicable to ski rental and job scheduling problems

Abstract

In this work we study the problem of using machine-learned predictions to improve the performance of online algorithms. We consider two classical problems, ski rental and non-clairvoyant job scheduling, and obtain new online algorithms that use predictions to make their decisions. These algorithms are oblivious to the performance of the predictor, improve with better predictions, but do not degrade much if the predictions are poor.