We consider the learnability of the quantum neural network (QNN) built on the variational hybrid quantum-classical scheme, which remains largely unknown due to the non-convex optimization landscape, the measurement error, and the unavoidable gate errors introduced by noisy intermediate-scale quantum (NISQ) machines. Our contributions in this paper are multi-fold. First, we derive the utility bounds of QNN towards empirical risk minimization, and show that large gate noise, few quantum measurements, and deep circuit depth will lead to the poor utility bounds. This result also applies to the variational quantum circuits with gradient-based classical optimization, and can be of independent interest. We then prove that QNN can be treated as a differentially private (DP) model. Thirdly, we show that if a concept class can be efficiently learned by QNN, then it can also be effectively learned by QNN even with gate noise. This result implies the same learnability of QNN whether it is implemented on noiseless or noisy quantum machines. We last exhibit that the quantum statistical query (QSQ) model can be effectively simulated by noisy QNN. Since the QSQ model can tackle certain tasks with runtime speedup, our result suggests that the modified QNN implemented on NISQ devices will retain the quantum advantage. Numerical simulations support the theoretical results.
Unifying Charge-Learnability Transitions in Monitored \(U(1)\)-Symmetric Quantum Circuits through Informational Power of Local Measurement
Key Laboratory of Atomic and Subatomic Structure and Quantum Control (Ministry of Education), Guangdong Basic Research Center of Excellence for Structure and Fundamental Interactions of Matter, and School of Physics, South China Normal University(原子与亚原子结构及量子控制重点实验室(教育部),广东省物质结构与基本相互作用卓越研究中心,华南师范大学物理学院);Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials, Guangdong-Hong Kong Joint Laboratory of Quantum Matter, and Frontier Research Institute for Physics, South China Normal University(广东省量子工程与量子材料重点实验室,粤港澳量子物质联合实验室,华南师范大学前沿物理研究所)
Learnability, Identifiability, and Monitor Placement in Quantum Network Tomography
CONNECT Research Centre, School of Engineering, Trinity College Dublin(都柏林三一学院工程学院 CONNECT 研究中心);Manning College of Information and Computer Science, University of Massachusetts, Amherst(马萨诸塞大学阿默斯特分校曼宁信息与计算机科学学院);Walton Institute, South East Technological University(东南理工大学沃尔顿研究所)
On the Quantum versus Classical Learnability of Discrete Distributions
Dahlem Center for Complex Quantum Systems, Freie Universität Berlin(柏林自由大学达勒姆复杂量子系统中心);TU Berlin(柏林工业大学);FhG SIT(弗劳恩霍夫系统与创新研究所);Helmholtz Center Berlin(柏林亥姆霍兹中心);Department of Mathematics and Computer Science, Freie Universität Berlin(柏林自由大学数学与计算机科学系)
Mixed-state learnability transitions in monitored noisy quantum dynamics
Max Planck Institute for the Physics of Complex Systems(马克斯·普朗克复杂系统物理研究所);University of Geneva(日内瓦大学);University of British Columbia(不列颠哥伦比亚大学);Princeton University(普林斯顿大学)
Learnability of the output distributions of local quantum circuits
Freie Universität Berlin(柏林自由大学);Helmholtz-Zentrum Berlin für Materialien und Energie(亥姆霍兹柏林材料与能源研究中心);Stanford University(斯坦福大学);Joint Center for Quantum Information and Computer Science (QuICS), University of Maryland(马里兰大学量子信息与计算机科学联合中心);NIST(美国国家标准与技术研究院);TU Berlin(柏林工业大学);FhG SIT(弗劳恩霍夫安全信息技术研究所);Fraunhofer Heinrich Hertz Institute(弗劳恩霍夫海因里希赫兹研究所)
From Reachability to Learnability: Geometric Design Principles for Quantum Neural Networks
Institute for Particle Physics Phenomenology, Department of Physics Durham University(粒子物理现象研究所,物理系杜伦大学);Institute for Theoretical Physics, Karlsruhe Institute of Technology (KIT)(理论物理研究所,卡尔斯鲁厄理工学院(KIT))
Data-Driven Learnability Transition of Measurement-Induced Entanglement
State Key Laboratory of Surface Physics(表面物理国家重点实验室);Department of Physics, Fudan University, Shanghai 200433, China(复旦大学物理系);Shanghai Research Center for Quantum Sciences, Shanghai 201315, China(上海量子科学研究中心);Institute for Nanoelectronic Devices(纳米电子器件研究所);Quantum Computing, Fudan University, Shanghai 200433, China(量子计算,复旦大学);Hefei National Laboratory, Hefei 230088, China(合肥国家实验室)
National Taiwan University(国立台湾大学);University of Technology Sydney(悉尼科技大学)
近年来,量子机器学习受到了极大的关注,在开发量子算法以加速传统机器学习任务方面取得了可喜的进展。然而,在这项工作中,我们专注于研究样本复杂度的信息论上限-多少训练样本足以预测未知目标函数的未来行为。可以说,这类问题是统计学习理论中最基本的问题之一,实际设置的界限可以完全由一个简单的复杂性度量来表征。我们的主要结果是,对于学习一个未知的量子测量,上界,由脂肪粉碎维数,是线性正比于底层希尔伯特空间的维数。学习一个未知的量子态成为我们的对偶问题,作为副产品,我们可以恢复Aaronson的著名结果[Proc. R. Soc. A 463:3089-3144(2007)]。此外,其他著名的复杂性措施,如覆盖数和Rademacher复杂性推导明确。我们能够将样品复杂性的测量与量子信息科学的各个领域联系起来,例如量子态/测量层析成像,量子态鉴别和量子随机存取码,这些领域可能是独立感兴趣的。最后,在一般布洛赫球表示的帮助下,我们证明了学习量子测量/状态可以在数学上表示为神经网络。因此,经典ML算法可以有效地完成这两个量子学习任务。
Statistical Mechanics of Non-Abelian Learnability Transitions
Kavli Institute for Theoretical Physics, University of California, Santa Barbara(加州大学圣塔芭芭拉分校卡弗里理论物理研究所);Department of Theoretical Physics, University of Geneva(日内瓦大学理论物理系);Geneva Quantum Center, University of Geneva(日内瓦大学日内瓦量子中心)
A Unified Frequency Principle for Quantum and Classical Machine Learning
Yau Mathematical Sciences Center, Tsinghua University(清华大学丘成桐数学科学中心);Department of Mathematics, Tsinghua University(清华大学数学系);Center for Quantum Information, IIIS, Tsinghua University(清华大学交叉信息研究院量子信息中心);Yanqi Lake Beijing Institute of Mathematical Sciences and Applications(北京雁栖湖应用数学研究院);Shanghai Qi Zhi Institute(上海人工智能实验室);Hefei National Laboratory(合肥国家实验室)
Quantum hardness of learning shallow classical circuits
Center for Theoretical Physics, MIT(麻省理工学院理论物理中心);CWI(CWI数学与计算机科学中心);QuSoft(QuSoft量子软件研究中心);Joint Center for Quantum Information and Computer Science, University of Maryland(马里兰大学量子信息与计算机科学联合中心)
本文研究了均匀分布和分布无关的PAC学习框架下等深度经典电路的量子可学性。为了实现我们的结果,我们建立了量子学习和量子安全密码系统之间的联系。然后,我们实现以下结果。1)在均匀分布下学习AC $^0$和TC $^0$的难度。我们的第一个结果涉及概念类TC $^0$(resp. AC $^0$),具有无界扇入多数门(分别为)的恒定深度和多项式大小的电路类。AND、OR、NOT门)。我们证明,如果不存在量子多项式时间(分别)。强次指数时间)算法来解决带误差的环学习(RLWE)问题,那么TC $^0$(分别为. AC $^0$)(即使访问量子成员资格查询)。该结果中的主要技术使用被认为是量子安全的显式伪随机函数来构造在均匀分布下难以量子学习的概念类。2)在PAC设置中学习TC $^0_2$的难度。我们的第二个结果表明,如果不存在LWE问题的量子多项式时间算法,则不存在TC类$^0_2$的多项式时间量子PAC学习算法,即,深度2 TC $^0$电路。在这个结果中的主要技术是建立一个连接的公钥密码系统的量子安全性和可学习的概念类,包括解密功能的密码系统。这对Aaronson [Aar 05]提出的“量子计算理论的十大半重大挑战”之一给出了强(条件)否定的答案,Aaronson询问AC $^0$和TC $^0$是否可以在量子多项式时间内进行PAC学习。
Provable learning separation for predicting time-evolution of quantum many-body systems
Leiden University(莱顿大学);Leiden Institute of Advanced Computer Science(莱顿高级计算机科学研究所);Dahlem Center for Complex Quantum Systems(达赫姆复杂量子系统中心);Fraunhofer Heinrich Hertz Institute(弗劳恩霍夫海因里希·赫兹研究所)
Statistical Analysis of Quantum State Learning Process in Quantum Neural Networks
Tsinghua University(清华大学);Institute for Advanced Study, Tsinghua University(清华大学高等研究院);Baidu Research(百度研究院);Institute for Quantum Computing, Baidu Research(百度研究院量子计算研究所);Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州))