Online Learning with Recency: Algorithms for Sliding-window Streaming Multi-armed Bandits
This paper studies algorithms for single-pass sliding-window streaming multi-armed bandits with limited memory.
The paper introduces a new model of single-pass sliding-window streaming multi-armed bandits with limited memory, which is a natural extension of the streaming multi-armed bandits model.
Before reading this…
Applications
- →Online learning
- →Decision-making under uncertainty
To understand this paper, make sure you know these concepts first:
- Multi-armed banditsfind papers →
- Streaming algorithmsfind papers →
Abstract
More Like ThisMotivated by the recency effect in online learning, we study algorithms for single-pass *sliding-window streaming multi-armed bandits (MABs)* in this paper. In this setting, we are given $n$ arms with unknown sub-Gaussian reward distributions and a parameter $W$. The arms arrive in a single-pass stream, and only the most recent $W$ arms are considered valid. The algorithm is required to perform pure exploration and regret minimization with limited memory, defined as the number of stored arms. The model is a natural extension of the streaming multi-armed bandits model (without the sliding window) that has been extensively studied in recent years. We provide a comprehensive analysis of both the pure exploration and regret minimization problems with the model. For pure exploration, we prove that finding the best arm is hard with sublinear memory while finding an approximate best arm admits an efficient algorithm. For regret minimization, we explore a new notion of regret and give sharp memory-regret trade-offs for any single-pass algorithm. We complement our theoretical results with experiments, demonstrating the trade-offs between sample, regret, and memory.