arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.05294cs.LG
收藏

结构化偏好反馈的纯探索

Pure Exploration with Structured Preference Feedback

Indian Institute of Science, Bangalore(印度科学理工学院,班加罗尔);Amazon, Palo Alto, CA(亚马逊,帕洛阿尔托)

Shubham Gupta, Aadirupa Saha, Sumeet Katariya

摘要:

我们考虑的问题,纯粹的探索与子集明智的偏好反馈,其中包含$N$武器的功能。允许学习者查询大小为$K$的子集,并以嘈杂的获胜者的形式接收反馈。学习器的目标是使用尽可能少的查询来有效地识别最佳手臂。这种设置在涉及人工反馈的各种在线决策场景中是相关的,例如在线零售、流媒体服务、新闻馈送和在线广告;因为人们从子集中选择首选项目比单独为项目分配喜爱度评分更容易和更可靠。据我们所知,这是第一个工作,认为子集明智的偏好反馈模型在一个结构化的设置,它允许潜在的无限集的武器。我们提出了两种算法,保证最好的手臂在$\tilde{O} (\frac{d^2}{K Δ^2})$样本的检测概率至少为$1 - δ$,其中$d$是手臂的功能和$Δ$之间的效用差距的适当的概念。我们还推导出了一个依赖于实例的$Ω(\frac{d}{Δ^2} \log \frac{1}δ)$的下限,它与我们在最坏情况下的上限相匹配。最后,我们进行了大量的实验来证实我们的理论发现,并观察到我们的自适应算法停止,需要的样本比非自适应算法少12倍。

英文摘要:

We consider the problem of pure exploration with subset-wise preference feedback, which contains $N$ arms with features. The learner is allowed to query subsets of size $K$ and receives feedback in the form of a noisy winner. The goal of the learner is to identify the best arm efficiently using as few queries as possible. This setting is relevant in various online decision-making scenarios involving human feedback such as online retailing, streaming services, news feed, and online advertising; since it is easier and more reliable for people to choose a preferred item from a subset than to assign a likability score to an item in isolation. To the best of our knowledge, this is the first work that considers the subset-wise preference feedback model in a structured setting, which allows for potentially infinite set of arms. We present two algorithms that guarantee the detection of the best-arm in $\tilde{O} (\frac{d^2}{K Δ^2})$ samples with probability at least $1 - δ$, where $d$ is the dimension of the arm-features and $Δ$ is the appropriate notion of utility gap among the arms. We also derive an instance-dependent lower bound of $Ω(\frac{d}{Δ^2} \log \frac{1}δ)$ which matches our upper bound on a worst-case instance. Finally, we run extensive experiments to corroborate our theoretical findings, and observe that our adaptive algorithm stops and requires up to 12x fewer samples than a non-adaptive algorithm.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑