面向含噪中规模量子处理器的量子机器学习  

Quantum Machine Learning with Noisy Intermediate-Scale Quantum Processors

作　　者：石金晶[1] 肖子萌 王雯萱 张师超[3] 李学龙 SHI Jin-Jing;XIAO Zi-Meng;WANG Wen-Xuan;ZHANG Shi-Chao;LI Xue-Long(School of Electronic Information,Central South University,Changsha 410004;School of Computer Science and Engineering,Central South University,Changsha 410083;Guangxi Key Laboratory of Multi-Source Information Mining and Security,Guangxi Normal University,Guilin,Guangxi 541004;Institute of Artificial Intelligence(Tele AI),China Telecom,Beijing 100033)

机构地区：[1]中南大学电子信息学院,长沙410004 [2]中南大学计算机学院,长沙410083 [3]广西师范大学广西多源信息挖掘与安全重点实验室,广西桂林541004 [4]中国电信人工智能研究院,北京100033

出　　处：《计算机学报》2025年第3期602-631,共30页Chinese Journal of Computers

基　　金：国家自然科学基金(62272483);湖南省自然科学基金杰出青年基金(2023JJ10078);湖南省研究生科研创新项目(CX20240266)资助。

摘　　要：量子计算与人工智能结合,在增强模型表达能力、加速和优化机器学习等方面可能产生颠覆性影响,有望突破人工智能领域所面临的可解释性差、最优解难等问题,量子人工智能已成为国内外重点关注的学科前沿。量子机器学习是量子人工智能领域的重要研究内容,它将量子计算基础理论与机器学习原理相结合,以实现具有量子加速的机器学习任务。随着量子计算软硬件的快速发展,含噪中规模量子(NISQ)处理器的学习优势被证明,国内外学者相继提出一系列量子机器学习方法,以挖掘量子计算助力人工智能技术发展的创新应用。然而,当前的量子机器学习仍局限于对算法的优化,缺乏系统层面的理论架构,仍有许多科学问题亟待解决。本文首先从量子机器学习系统表征角度出发,建立量子机器学习系统的层次模型,概括和总结了面向各类任务的量子机器学习方案,分析了量子机器学习在提高经典算法速度等方面可能体现的“量子优势”。接着根据量子机器学习系统的层次结构,从原理层、计算层、应用层这三个方面对现有量子机器学习方法进行了总结与梳理,系统性地分析和讨论了其中的关键问题与解决方案。最后,结合当前阶段量子人工智能的发展特点,重点分析了量子机器学习领域面临的科学问题与挑战,并对未来该领域的发展趋势进行了深入分析与展望。Quantum computing is a new type of computing model that follows the laws of quantum mechanics.By leveraging the characteristics of quantum entanglement and superposition,it can theoretically achieve exponential acceleration of classical algorithms.The combination of quantum computing and artificial intelligence may have a transformative impact on enhancing model representation ability,accelerating and optimizing machine learning.It is expected to break through the problems of poor interpretability and difficulty deriving optimal solutions in the field of artificial intelligence.Quantum artificial intelligence has increasingly become a technological highland that countries around the world compete for.More and more researchers have begun to explore a breakthrough to help the development of artificial intelligence through quantum computing.Quantum machine learning is an important research area in the field of quantum artificial intelligence,which combines the basic theory of quantum computing with the principles of machine learning.It is expected to use the potential'quantum advantage'to break through the problems that are difficult to solve by classical algorithms and improve the computational efficiency of machine learning models.With the rapid development of quantum computing hardware and software,the learning advantages of noisy intermediate-scale quantum(NISQ)processors have been proved.Scholars both domestically and internationally have successively proposed a series of quantum machine learning methods to explore the innovative applications of quantum computing to help the development of artificial intelligence technology.In the future,with the fault tolerance,generalization,and large-scale development of quantum computing devices,the'quantum advantage'of quantum machine learning is expected to be fully explored,and then applied to engineering and industry,bringing new transformations to the field of artificial intelligence.However,quantum machine learning is still limited to algorithm optimization and lacks

关 键 词：量子计算 量子人工智能 量子机器学习 量子算法 含噪中规模量子处理器 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]