标题： 量子属性后门（QuPTs）用于攻击量子神经网络 显示英文标题

标题： Quantum Properties Trojans (QuPTs) for Attacking Quantum Neural Networks

摘要： 量子神经网络（QNN）在量子机器学习（QML）的未来具有巨大的潜力。 然而，QNN的安全性和鲁棒性仍 largely 未被探索。 在本工作中，我们提出了一种基于量子计算特性的新型后门攻击，在基于QNN的二分类器中进行攻击。 我们提出的量子特性后门（QuPTs）基于量子门的单位性质，插入噪声和哈达玛门以实现叠加，从而开发后门并攻击QNN。 我们展示了所提出的QuPTs具有显著的隐蔽性，并对量子电路的性能产生严重影响，特别是QNN。 最有效的QuPT在实验设置下导致受损QNN的准确率下降了23%。 据我们所知，这是第一项针对完全量子神经网络的后门攻击研究，独立于任何混合经典-量子架构。 显示英文摘要

摘要： Quantum neural networks (QNN) hold immense potential for the future of quantum machine learning (QML). However, QNN security and robustness remain largely unexplored. In this work, we proposed novel Trojan attacks based on the quantum computing properties in a QNN-based binary classifier. Our proposed Quantum Properties Trojans (QuPTs) are based on the unitary property of quantum gates to insert noise and Hadamard gates to enable superposition to develop Trojans and attack QNNs. We showed that the proposed QuPTs are significantly stealthier and heavily impact the quantum circuits' performance, specifically QNNs. The most impactful QuPT caused a deterioration of 23% accuracy of the compromised QNN under the experimental setup. To the best of our knowledge, this is the first work on the Trojan attack on a fully quantum neural network independent of any hybrid classical-quantum architecture.