arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.08549cs.LGcs.DSstat.ML
收藏

Oracle高效的在线学习,超越最坏情况的对手

Oracle-Efficient Online Learning for Beyond Worst-Case Adversaries

University of California, Berkeley(加州大学伯克利分校)

Nika Haghtalab, Yanjun Han, Abhishek Shetty, Kunhe Yang

摘要:

在本文中,我们研究了甲骨文有效的算法,超越最坏情况分析的在线学习。我们专注于两个设置。首先,[RST 11,HRS 22]的平滑分析设置,其中对手被约束为从密度上限为$1/σ$乘以均匀密度的分布生成样本。第二,设置$K$ -提示转换学习,其中学习者被给予每个时间步长对$K$提示的访问,所述提示被保证包括真实实例。我们给出了第一个已知的甲骨文有效的算法,这两个设置只依赖于伪(或VC)维的类和参数$σ$和$K$,捕捉对手的力量。特别是,我们实现了$ \widetilde{O} ( \sqrt{T dσ^{-1}} ) $和$ \widetilde{O} ( \sqrt{T dK} ) $学习实值函数和$ O ( \sqrt{T dσ^{-\frac{1}{2}} } )$学习二进制值函数的预言有效的遗憾界。对于平滑的分析设置,我们的结果给出了第一个具有平滑对手的在线学习的预言机有效算法[HRS 22]。这对比了[HK 16]建立的在线学习与最坏情况对手和离线学习之间的计算分离。我们的算法还实现了改进的界限,最坏情况下设置小域。特别地,我们给出了一个预言机有效的算法,遗憾的是$O ( \sqrt{T(d |\mathcal{X}|)^{1/2} })$,这是一个改进的早期$O ( \sqrt{T|\mathcal{X}|})$绑定[DS 16]。

英文摘要:

In this paper, we study oracle-efficient algorithms for beyond worst-case analysis of online learning. We focus on two settings. First, the smoothed analysis setting of [RST11,HRS22] where an adversary is constrained to generating samples from distributions whose density is upper bounded by $1/σ$ times the uniform density. Second, the setting of $K$-hint transductive learning, where the learner is given access to $K$ hints per time step that are guaranteed to include the true instance. We give the first known oracle-efficient algorithms for both settings that depend only on the pseudo (or VC) dimension of the class and parameters $σ$ and $K$ that capture the power of the adversary. In particular, we achieve oracle-efficient regret bounds of $ \widetilde{O} ( \sqrt{T dσ^{-1}} ) $ and $ \widetilde{O} ( \sqrt{T dK} ) $ for learning real-valued functions and $ O ( \sqrt{T dσ^{-\frac{1}{2}} } )$ for learning binary-valued functions. For the smoothed analysis setting, our results give the first oracle-efficient algorithm for online learning with smoothed adversaries [HRS22]. This contrasts the computational separation between online learning with worst-case adversaries and offline learning established by [HK16]. Our algorithms also achieve improved bounds for worst-case setting with small domains. In particular, we give an oracle-efficient algorithm with regret of $O ( \sqrt{T(d |\mathcal{X}|)^{1/2} })$, which is a refinement of the earlier $O ( \sqrt{T|\mathcal{X}|})$ bound by [DS16].

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。

↑