论文标题
拜占庭式联合线性土匪
Byzantine-Robust Federated Linear Bandits
论文作者
论文摘要
在本文中，我们在联合环境中研究了一个线性匪徒优化问题，其中大量分布式代理会协作学习常见的线性匪徒模型。适用于此设置的标准联合学习算法也容易受到拜占庭式攻击，即使是一小部分代理。我们提出了一种具有强大的聚合甲骨文的新型算法，该算法利用了几何中位数。我们证明，我们提出的算法对拜占庭的攻击是强大的，对少数不到一半的代理商，并且可以实现$ \ tilde {\ Mathcal {o}}}}（{t^{3/4}}} $遗憾的是，$ \ \ \ \ \ \ \ \ \ \ \ \ \ {o}（o}（o）此外，我们通过基于树的机制使我们的算法差异化。最后，如果已知腐败的水平很小，我们表明，使用Mean Oracle的几何中位数进行健壮的聚合进一步改善了遗憾。
In this paper, we study a linear bandit optimization problem in a federated setting where a large collection of distributed agents collaboratively learn a common linear bandit model. Standard federated learning algorithms applied to this setting are vulnerable to Byzantine attacks on even a small fraction of agents. We propose a novel algorithm with a robust aggregation oracle that utilizes the geometric median. We prove that our proposed algorithm is robust to Byzantine attacks on fewer than half of agents and achieves a sublinear $\tilde{\mathcal{O}}({T^{3/4}})$ regret with $\mathcal{O}(\sqrt{T})$ steps of communication in $T$ steps. Moreover, we make our algorithm differentially private via a tree-based mechanism. Finally, if the level of corruption is known to be small, we show that using the geometric median of mean oracle for robust aggregation further improves the regret bound.
