Simultaneously Achieving Group Exposure Fairness and Within-Group Meritocracy in Stochastic Bandits
作者: Subham Pokhriyal, Shweta Jain, Ganesh Ghalme, Swapnil Dhamal, Sujit Gujar
分类: cs.LG, cs.AI, cs.CY, cs.MA
发布日期: 2024-02-08
备注: Accepted in AAMAS 2024
💡 一句话要点
提出双层公平性以解决随机多臂老虎机中的公平性问题
🎯 匹配领域: 支柱一:机器人控制 (Robot Control)
关键词: 随机多臂老虎机 公平性 双层公平性 优点公平性 UCB算法 资源分配 在线广告
📋 核心要点
- 现有方法主要关注单个臂的曝光保证,未能有效处理臂的分组情况,导致公平性不足。
- 本文提出双层公平性,确保每个组的最低曝光量,并在组内实现基于优点的公平性。
- 实验结果表明,BF-UCB算法在后悔值上实现了次线性增长,且在曝光保证方面优于现有算法。
📝 摘要(中文)
现有的随机多臂老虎机(MAB)公平性方法主要关注单个臂的曝光保证。当臂根据某些属性自然分组时,本文提出了双层公平性(Bi-Level Fairness),考虑了两个层次的公平性。在第一个层次上,双层公平性保证每个组的最低曝光量。为了解决组内单个臂的拉取不平衡问题,第二个层次引入了基于优点的公平性,确保每个臂根据其在组内的优点被拉取。我们的工作表明,可以通过适应基于UCB的算法来实现双层公平性,提供随时的组曝光公平性保证,并确保组内的个体优点公平性。通过模拟实验,我们进一步展示了BF-UCB算法在后悔值上实现了次线性增长,并在组和个体曝光保证方面优于现有算法。
🔬 方法详解
问题定义:本文解决随机多臂老虎机中的公平性问题,现有方法未能有效处理臂的分组情况,导致曝光不均和优点不平衡的问题。
核心思路:提出双层公平性,第一层保证每个组的最低曝光量,第二层确保组内每个臂根据其优点被拉取,从而实现公平性与效益的平衡。
技术框架:算法BF-UCB基于UCB框架,分为两个主要模块:组曝光公平性模块和优点公平性模块,分别处理组间和组内的公平性问题。
关键创新:最重要的创新在于将公平性分为两个层次,首次在随机多臂老虎机中实现了组曝光和优点公平性的同时保证,显著提升了公平性效果。
关键设计:BF-UCB算法通过优化拉取策略,平衡组曝光公平性和优点公平性,确保后悔值的上限为O(√T),并通过参数调节实现最佳性能。
🖼️ 关键图片
📊 实验亮点
实验结果显示,BF-UCB算法在后悔值上实现了次线性增长,优于现有算法,并且在组和个体曝光保证方面提供了更好的结果。具体而言,BF-UCB在不显著降低奖励的情况下,达到了O(√T)的后悔值上限,展现了其优越性。
🎯 应用场景
该研究的潜在应用领域包括在线广告投放、推荐系统和资源分配等场景,能够有效提升不同用户群体的公平性和满意度。未来,该方法有望在多种需要公平性考量的决策系统中得到广泛应用,推动相关领域的研究与实践。
📄 摘要(原文)
Existing approaches to fairness in stochastic multi-armed bandits (MAB) primarily focus on exposure guarantee to individual arms. When arms are naturally grouped by certain attribute(s), we propose Bi-Level Fairness, which considers two levels of fairness. At the first level, Bi-Level Fairness guarantees a certain minimum exposure to each group. To address the unbalanced allocation of pulls to individual arms within a group, we consider meritocratic fairness at the second level, which ensures that each arm is pulled according to its merit within the group. Our work shows that we can adapt a UCB-based algorithm to achieve a Bi-Level Fairness by providing (i) anytime Group Exposure Fairness guarantees and (ii) ensuring individual-level Meritocratic Fairness within each group. We first show that one can decompose regret bounds into two components: (a) regret due to anytime group exposure fairness and (b) regret due to meritocratic fairness within each group. Our proposed algorithm BF-UCB balances these two regrets optimally to achieve the upper bound of $O(\sqrt{T})$ on regret; $T$ being the stopping time. With the help of simulated experiments, we further show that BF-UCB achieves sub-linear regret; provides better group and individual exposure guarantees compared to existing algorithms; and does not result in a significant drop in reward with respect to UCB algorithm, which does not impose any fairness constraint.