arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.03365stat.MLcs.LG
收藏

具有局部差分隐私的广义线性Bandits

Generalized Linear Bandits with Local Differential Privacy

The Hong Kong University of Science and Technology(香港科技大学)

Yuxuan Han, Zhipeng Liang, Yang Wang, Jiheng Zhang

摘要:

上下文强盗算法在个性化在线决策中是有用的。然而,许多应用程序,如个性化医疗和在线广告需要利用个人特定的信息进行有效的学习,而用户的数据应该保持私有的服务器由于隐私问题。这促使引入局部差分隐私(LDP),一个严格的隐私概念,上下文强盗。在本文中,我们设计了随机广义线性强盗LDP算法,以实现在非隐私设置相同的遗憾界。我们的主要思想是开发一个随机的基于梯度的估计和更新机制,以确保LDP。然后,我们利用随机梯度下降(SGD)的灵活性,其理论保证土匪问题很少探索,在处理广义线性土匪。我们还开发了一个估计器和更新机制的基础上,普通最小二乘(OLS)的线性土匪。最后,我们进行了仿真和真实世界的数据集的实验,以证明我们的算法在LDP约束下具有合理的小参数$(\varepsilon, δ)$,以确保强大的隐私保护的一贯高超的性能。

英文摘要:

Contextual bandit algorithms are useful in personalized online decision-making. However, many applications such as personalized medicine and online advertising require the utilization of individual-specific information for effective learning, while user's data should remain private from the server due to privacy concerns. This motivates the introduction of local differential privacy (LDP), a stringent notion in privacy, to contextual bandits. In this paper, we design LDP algorithms for stochastic generalized linear bandits to achieve the same regret bound as in non-privacy settings. Our main idea is to develop a stochastic gradient-based estimator and update mechanism to ensure LDP. We then exploit the flexibility of stochastic gradient descent (SGD), whose theoretical guarantee for bandit problems is rarely explored, in dealing with generalized linear bandits. We also develop an estimator and update mechanism based on Ordinary Least Square (OLS) for linear bandits. Finally, we conduct experiments with both simulation and real-world datasets to demonstrate the consistently superb performance of our algorithms under LDP constraints with reasonably small parameters $(\varepsilon, δ)$ to ensure strong privacy protection.

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。

↑