arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2106.06469cs.LGcs.CG
收藏

木马神经网络的拓扑检测

Topological Detection of Trojaned Neural Networks

Stony Brook University(石溪大学);Morgan Stanley(摩根士丹利);IST Austria(奥地利科学与技术研究所);City University of New York(纽约城市大学)

Songzhu Zheng, Yikai Zhang, Hubert Wagner, Mayank Goswami, Chao Chen

摘要:

众所周知,深度神经网络存在安全问题。一个特别的威胁是特洛伊木马攻击。当攻击者通过特洛伊木马的训练样本悄悄地操纵模型的行为时,就会发生这种情况,这些样本以后可以被利用。在基本神经科学原理的指导下,我们发现了特洛伊模型特征的微妙但关键的结构偏差。在我们的分析中,我们使用拓扑工具。它们使我们能够对网络中的高阶依赖关系进行建模,鲁棒地比较不同的网络,并定位结构异常。一个有趣的观察是,木马模型开发了从输入到输出层的捷径。受这些观察的启发,我们设计了一个策略,木马模型的鲁棒检测。与标准基线相比,它在多个基准测试中显示出更好的性能。

英文摘要:

Deep neural networks are known to have security issues. One particular threat is the Trojan attack. It occurs when the attackers stealthily manipulate the model's behavior through Trojaned training samples, which can later be exploited. Guided by basic neuroscientific principles we discover subtle -- yet critical -- structural deviation characterizing Trojaned models. In our analysis we use topological tools. They allow us to model high-order dependencies in the networks, robustly compare different networks, and localize structural abnormalities. One interesting observation is that Trojaned models develop short-cuts from input to output layers. Inspired by these observations, we devise a strategy for robust detection of Trojaned models. Compared to standard baselines it displays better performance on multiple benchmarks.

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。

↑