超导量子计算机中基于测量的串扰的安全影响

中文摘要 AI 辅助

串扰是超导量子处理器可扩展性的主要问题之一。但它也可能被威胁行为者利用，故意破坏计算和窃取信息。本文分析了量子比特测量过程中因串扰导致的数据泄露。先前的研究在IBM量子计算机上对单个量子比特的测量进行分类时，达到了96%的准确率。当在VTT的超导硬件上评估时，我们使用类似的框架达到了71.68%的准确率。我们扩展的框架添加了额外标签，在使用支持向量机（SVM）对单个量子比特的测量进行分类时，准确率达到76.30%，对两个量子比特的测量进行分类时，准确率达到53.37%。实验结果表明，对一个量子比特的测量操作会影响测量相邻量子比特的概率。对目标量子比特的影响取决于被测量的相邻量子比特的数量以及测量结果。

英文摘要

Crosstalk is one of the major issues for scalability of superconducting quantum processors. But it can also be used by threat actors to intentionally sabotage computations and steal information. This paper analyzes data leakage due to crosstalk during qubit measurement. Prior research yielded an accuracy of 96% in classifying measurement of one qubit in IBM quantum computers. When evaluated on superconducting hardware from VTT, we achieved an accuracy of 71.68% with similar framework. Our extended framework with additional labels yields an accuracy of 76.30% for classifying measurement of one qubit and 53.37% for measurement of two qubits using Support Vector Machine (SVM). Experimental results indicate that measurement operation on a qubit can affect the probability of measuring adjacent qubits. The impact on the target qubit is dependent on the number of neighboring qubits measured and also the measurement outcome.

发表机构

• University of Oulu（奥卢大学）

机构由 AI 辅助整理，请以论文原文为准。