在资源有限下高效找出最优选项，兼顾成本与准确性
Learning with a Budget: Identifying the Best Arm with Resource Constraints
- 提出资源感知的逐次减半算法，动态分配有限资源
- 理论证明在随机与确定性消耗下均能快速收敛
- 适合资源敏感场景，如医疗试验、算力受限实验
在诸多应用中，评估不同方案的效果伴随着不同的成本或资源消耗。针对这种异质性，我们研究了带有资源约束的最佳臂识别（BAIwRC）问题：智能体需在资源受限条件下识别最优选项（即最佳臂）。每次抽取臂会消耗一种或多种有限资源。本文提出两种关键贡献：首先，设计了资源配额制的逐次减半算法（SH-RR），将资源感知机制融入经典的逐次减半框架，统一处理随机与确定性资源消耗场景；其次，引入新的有效消耗度量，建立了完备的理论分析基础，确保在给定资源预算下以高概率识别最优臂。
原文摘要 · Abstract (English)
In many applications, evaluating the effectiveness of different alternatives comes with varying costs or resource usage. Motivated by such heterogeneity, we study the Best Arm Identification with Resource Constraints (BAIwRC) problem, where an agent seeks to identify the best alternative (aka arm) in the presence of resource constraints. Each arm pull consumes one or more types of limited resources. We make two key contributions. First, we propose the Successive Halving with Resource Rationing (SH-RR) algorithm, which integrates resource-aware allocation into the classical successive halving framework on best arm identification. The SH-RR algorithm unifies the theoretical analysis for both the stochastic and deterministic consumption settings, with a new \textit{effective consumption measure
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。