The pair-matching problem appears in many applications where one wants to discover good matches between pairs of entities or individuals. Formally, the set of individuals is represented by the nodes of a graph where the edges, unobserved at first, represent the good matches. The algorithm queries pairs of nodes and observes the presence/absence of edges. Its goal is to discover as many edges as possible with a fixed budget of queries. Pair-matching is a particular instance of multi-armed bandit problem in which the arms are pairs of individuals and the rewards are edges linking these pairs. This bandit problem is non-standard though, as each arm can only be played once. Given this last constraint, sublinear regret can be expected only if the graph presents some underlying structure. This paper shows that sublinear regret is achievable in the case where the graph is generated according to a Stochastic Block Model (SBM) with two communities. Optimal regret bounds are computed for this pair-matching problem. They exhibit a phase transition related to the Kesten-Stigum threshold for community detection in SBM. The pair-matching problem is considered in the case where each node is constrained to be sampled less than a given amount of times. We show how optimal regret rates depend on this constraint. The paper is concluded by a conjecture regarding the optimal regret when the number of communities is larger than 2. Contrary to the two communities case, we argue that a statistical-computational gap would appear in this problem.
翻译:配对问题出现在许多应用程序中, 人们想要发现对实体或个人的对对匹配。 形式上, 一组个人由图表的节点代表, 该图的边缘最初未观测到, 最初显示的边缘代表了好匹配。 算法查询对节点, 并观察边点的存在/ 没有边点。 目标是通过固定的查询预算来发现尽可能多的边缘。 配对是多臂强盗问题的特例, 双臂是个人配对的, 奖赏是连接这些对子的边缘。 这个突扰问题是非标准性的, 因为每个手臂只能播放一次。 鉴于此限制, 亚线性遗憾只有在图表显示某种基本结构时才能被期待。 本文显示, 亚线性遗憾在图形根据一个托查区块模型( SBM) 生成时是可以实现的。 最佳的遗憾界限是这个对配对问题。 他们展示了与 Kesten- Stigmm 界界点点的相向一个阶段过渡过渡, 因为每个社区检测的临界点在 SBM 中, 最优度中, 将显示的硬度限制程度。 。 我们的难度将显示, 最优的难度将显示的是, 最优程度的难度是最优的 。