๐ค AI Summary
This paper studies the prediction-augmented online caching problem, where each request is accompanied by a prediction of the next occurrence time of the corresponding page. Methodologically, we first improve the upper bound on the competitive ratio of the BlindOracle algorithm. Second, we establish, for the first time, a lower bound of ฮฉ(โh) on the competitive ratio of any randomized online caching algorithm, where h denotes the prediction error boundโthereby revealing the inherent hardness of the problem. Third, we propose a hybrid strategy combining BlindOracle and Marker, achieving an O(1)-competitive ratio under bounded prediction error, which is optimal up to constant factors. Collectively, our results significantly tighten both the upper and lower bounds on the competitive ratio for prediction-augmented caching, yielding the strongest known theoretical guarantees to date. This work provides a foundational benchmark for learning-augmented online algorithms.
๐ Abstract
We address the problem of learning-augmented online caching in the scenario when each request is accompanied by a prediction of the next occurrence of the requested page. We improve currently known bounds on the competitive ratio of the BlindOracle algorithm, which evicts a page predicted to be requested last. We also prove a lower bound on the competitive ratio of any randomized algorithm and show that a combination of the BlindOracle with the Marker algorithm achieves a competitive ratio that is optimal up to some constant.