Impact of Representation Learning in Linear Bandits

Impact of Representation Learning in Linear Bandits
复制标题

表征学习对线性强盗的影响

DOI:
--
复制
发表时间:
2020
期刊:
International Conference on Learning Representations
影响因子:
--
通讯作者:
S. Du
S. Du
中科院分区:
--
文献类型:
--
作者:
Jiaqi Yang;Wei Hu;Jason D. Lee;S. Du

文献摘要

参考文献

被引文献

相似文献

我们研究表示学习如何提高强盗问题的效率。我们研究了与尺寸$ d $同时播放$ t $线性土匪的设置,这些$ t $ bandit任务共享一个常见的$ k(ll d)$尺寸线性表示。对于有限行动设置,我们提出了一种新的算法,该算法可以实现$ widetilde {o}(tsqrt {kn} + sqrt {dknt})$遗憾,其中$ n $是我们为每个bandit演奏的回合数。当$ t $足够大时,我们的算法显着优于幼稚算法(独立播放$ t $ bandits),可以实现$ widetilde {o}(tsqrt {d n})$遗憾。我们还提供了$ OMEGA(TSQRT {KN} + SQRT {DKNT})$遗憾的下限,这表明我们的算法是最小值 - 最截至多层因素。此外,我们将我们的算法扩展到无限效法设置,并获得相应的遗憾界限,以证明在某些制度中的代表性学习的好处。我们还介绍了关于合成和现实世界数据的实验,以说明我们的理论发现并证明我们提出的算法的有效性。
We study how representation learning can improve the efficiency of bandit problems. We study the setting where we play $T$ linear bandits with dimension $d$ concurrently, and these $T$ bandit tasks share a common $k (ll d)$ dimensional linear representation. For the finite-action setting, we present a new algorithm which achieves $widetilde{O}(Tsqrt{kN} + sqrt{dkNT})$ regret, where $N$ is the number of rounds we play for each bandit. When $T$ is sufficiently large, our algorithm significantly outperforms the naive algorithm (playing $T$ bandits independently) that achieves $widetilde{O}(Tsqrt{d N})$ regret. We also provide an $Omega(Tsqrt{kN} + sqrt{dkNT})$ regret lower bound, showing that our algorithm is minimax-optimal up to poly-logarithmic factors. Furthermore, we extend our algorithm to the infinite-action setting and obtain a corresponding regret bound which demonstrates the benefit of representation learning in certain regimes. We also present experiments on synthetic and real-world data to illustrate our theoretical findings and demonstrate the effectiveness of our proposed algorithms.
DOI: --
发表时间: 2019-06
期刊: --
影响因子: --
作者:
M. Khodak;Maria-Florina Balcan;Ameet Talwalkar
通讯作者: M. Khodak;Maria-Florina Balcan;Ameet Talwalkar
无限武装线性上下文强盗的严格遗憾界限
DOI: --
发表时间: 2021
期刊: Proceedings of The 24th International Conference on Artificial Intelligence and Statistics
影响因子: --
作者:
Li, Yingkai;Wang, Yining;Chen, Xi;Zhou, Yuan
通讯作者: Zhou, Yuan
具有有限适应性和学习分布优化设计的线性老虎机
DOI: 10.1145/3406325.3451004
发表时间: 2021
期刊: STOC 2021: Proceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing
影响因子: --
作者:
Ruan, Yufei;Yang, Jiaqi;Zhou, Yuan
通讯作者: Zhou, Yuan