arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1603.06352cs.LG
收藏

低级别专家的在线学习

Online Learning with Low Rank Experts

Princeton University(普林斯顿大学);Microsoft Research(微软研究院);Technion—Israel Institute of Technology(以色列理工学院);The Hebrew University of Jerusalem(希伯来大学);Tel Aviv University(特拉维夫大学)

Elad Hazan, Tomer Koren, Roi Livni, Yishay Mansour

摘要:

我们考虑专家意见的预测问题时,专家的损失有低维结构:他们被限制到一个未知的$d$维子空间。我们设计的算法与遗憾界是独立的专家的数量,只依赖于排名$d$。对于随机模型,我们证明了$Θ(\sqrt{dT})$的一个紧界,并将其推广到一个近似$d$子空间的情形。对于对抗模型,我们给出了$O(d\sqrt{T})$的上界和$Ω(\sqrt{dT})$的下界。

英文摘要:

We consider the problem of prediction with expert advice when the losses of the experts have low-dimensional structure: they are restricted to an unknown $d$-dimensional subspace. We devise algorithms with regret bounds that are independent of the number of experts and depend only on the rank $d$. For the stochastic model we show a tight bound of $Θ(\sqrt{dT})$, and extend it to a setting of an approximate $d$ subspace. For the adversarial model we show an upper bound of $O(d\sqrt{T})$ and a lower bound of $Ω(\sqrt{dT})$.

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。

↑