Optimal Robustness-Consistency Trade-offs for Learning-Augmented Online Algorithms
This work addresses the fundamental trade-off between consistency and robustness in learning-augmented online algorithms, which is crucial for improving algorithm performance in real-world applications where predictions may be imperfect, though it is incremental as it builds on prior studies.
The paper tackles the problem of designing online algorithms that use machine-learned predictions to balance performance when predictions are accurate (consistency) with worst-case guarantees (robustness), by providing the first non-trivial lower bounds for competitive analysis in this context. It focuses on classic problems like ski-rental and non-clairvoyant scheduling, establishing optimal trade-offs in various settings.
We study the problem of improving the performance of online algorithms by incorporating machine-learned predictions. The goal is to design algorithms that are both consistent and robust, meaning that the algorithm performs well when predictions are accurate and maintains worst-case guarantees. Such algorithms have been studied in a recent line of works due to Lykouris and Vassilvitskii (ICML '18) and Purohit et al (NeurIPS '18). They provide robustness-consistency trade-offs for a variety of online problems. However, they leave open the question of whether these trade-offs are tight, i.e., to what extent to such trade-offs are necessary. In this paper, we provide the first set of non-trivial lower bounds for competitive analysis using machine-learned predictions. We focus on the classic problems of ski-rental and non-clairvoyant scheduling and provide optimal trade-offs in various settings.