维度的毒害
The poison of dimensionality
AI总结:
本文研究了模型规模对其易受投毒攻击脆弱性的影响,证明了在鲁棒梯度聚合规则下,参数量满足特定条件的线性与逻辑回归可被投毒者任意操纵,并揭示了模型表达力与攻击面之间的基本权衡。
摘要:
本文推进了对机器学习模型规模如何影响其易受投毒攻击脆弱性的理解,尽管存在最先进的防御机制。给定各向同性随机诚实特征向量,并将几何中位数(或截断均值)作为鲁棒梯度聚合规则,我们本质上证明了,或许令人惊讶的是,具有 $D \geq 169 H^2/P^2$ 个参数的线性回归和逻辑回归会受到投毒者的任意模型操纵,其中 $H$ 和 $P$ 是用于训练的诚实标记数据点和投毒数据点的数量。我们的实验进一步揭示了在合成数据,以及在 MNIST 和 FashionMNIST 数据上使用随机特征的线性分类器中,增强模型表达力与扩大投毒者攻击面之间的基本权衡。我们还讨论了对基于源的学习和神经网络的潜在影响。
英文摘要:
This paper advances the understanding of how the size of a machine learning model affects its vulnerability to poisoning, despite state-of-the-art defenses. Given isotropic random honest feature vectors and the geometric median (or clipped mean) as the robust gradient aggregator rule, we essentially prove that, perhaps surprisingly, linear and logistic regressions with $D \geq 169 H^2/P^2$ parameters are subject to arbitrary model manipulation by poisoners, where $H$ and $P$ are the numbers of honestly labeled and poisoned data points used for training. Our experiments go on exposing a fundamental tradeoff between augmenting model expressivity and increasing the poisoners' attack surface, on both synthetic data, and on MNIST & FashionMNIST data for linear classifiers with random features. We also discuss potential implications for source-based learning and neural nets.
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。