arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1903.00543cs.LGstat.ML
收藏

具有相对反馈的组合Bandits

Combinatorial Bandits with Relative Feedback

Indian Institute of Science(印度科学研究所)

Aadirupa Saha, Aditya Gopalan

摘要:

我们考虑组合在线学习与子集的选择时,只有相对反馈信息的子集是可用的,而不是强盗或半强盗反馈,这是绝对的。具体而言,我们研究了两个遗憾最小化问题的子集的有限基集$[n]$,与子集明智的相对偏好信息反馈根据多项logit选择模型。在第一设置中,学习者可以播放由最大大小限定的大小的子集,并且接收顶部$m$等级排序的反馈,而在第二设置中,学习者可以播放固定大小的子集$k$,其中观察到完整的子集等级作为反馈。对于这两种设置,我们设计的实例依赖和订单最优后悔算法与遗憾$O(\frac{n}{m} \ln T)$和$O(\frac{n}{k} \ln T)$,分别。我们得出的基本限制的遗憾性能的在线学习与子集明智的喜好,证明了我们的遗憾保证的紧密性。我们的研究结果还表明,引发更一般的顶部$m$排名排序反馈的价值超过单一赢家反馈($m=1$)。我们的理论结果与实证评估相印证。

英文摘要:

We consider combinatorial online learning with subset choices when only relative feedback information from subsets is available, instead of bandit or semi-bandit feedback which is absolute. Specifically, we study two regret minimisation problems over subsets of a finite ground set $[n]$, with subset-wise relative preference information feedback according to the Multinomial logit choice model. In the first setting, the learner can play subsets of size bounded by a maximum size and receives top-$m$ rank-ordered feedback, while in the second setting the learner can play subsets of a fixed size $k$ with a full subset ranking observed as feedback. For both settings, we devise instance-dependent and order-optimal regret algorithms with regret $O(\frac{n}{m} \ln T)$ and $O(\frac{n}{k} \ln T)$, respectively. We derive fundamental limits on the regret performance of online learning with subset-wise preferences, proving the tightness of our regret guarantees. Our results also show the value of eliciting more general top-$m$ rank-ordered feedback over single winner feedback ($m=1$). Our theoretical results are corroborated with empirical evaluations.

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。

↑