基于矩阵乘积算符的混合量子压缩经典生成对抗网络

doi:10.11896/jsjkx.240500017

计算机科学 ›› 2025, Vol. 52 ›› Issue (6): 74-81.doi: 10.11896/jsjkx.240500017

基于矩阵乘积算符的混合量子压缩经典生成对抗网络

张曜麟^1,2, 刘晓楠¹, 杜帅岐^1,2, 廉德萌^1,2

1 国家超级计算郑州中心郑州 450001
2 郑州大学计算机与人工智能学院郑州 450001

收稿日期:2024-05-06 修回日期:2024-08-26 出版日期:2025-06-15 发布日期:2025-06-11
通讯作者: 刘晓楠(prof.liu.xn@foxmail.com)
作者简介:(2474995768@qq.com)
基金资助:
2022年度河南省重大科技专项(221100210600)

Hybrid Quantum-classical Compressed Generative Adversarial Networks Based on Matrix Product Operators

ZHANG Yaolin^1,2, LIU Xiaonan¹, DU Shuaiqi^1,2, LIAN Demeng^1,2

1 National Supercomputing Center in Zhengzhou,Zhengzhou 450001,China
2 School of Computer and Artificial Intelligence,Zhengzhou University,Zhengzhou 450001,China

Received:2024-05-06 Revised:2024-08-26 Online:2025-06-15 Published:2025-06-11
About author:ZHANG Yaolin,born in 2000,postgra-duate.His main research interests include quantum algorithm and quantum machine learning.
LIU Xiaonan,born in 1977,Ph.D,associate,professor,master's supervisor.His main research interests include quantum algorithm and high-perfor-mance parallel computation.
Supported by:
Major Technology Project of Henan Province,China in 2022(221100210600).

摘要/Abstract

摘要： 神经网络在人工智能图像生成领域研究中占据重要地位。生成对抗网络作为近年来的热门算法,在图像生成任务中展现了卓越性能。量子计算作为一种新型计算模式,正在与传统人工智能算法融合,这样不仅加快了处理速度,还提升了数据安全性,尤其适合处理高维数据和优化问题。其中,混合量子经典生成对抗网络在图像生成任务中表现良好。然而,当前的混合量子经典生成模型在生成高维图像方面存在挑战,且生成器中线性层的加入,导致模型参数量增多。因此,提出了一种采用矩阵乘积算符的混合量子压缩经典生成对抗网络模型。该模型通过改进分块量子生成器的结构,使单次调用能够生成多个数据块,提高了模型效率。同时,结合经典网络的非线性特性和矩阵乘积算符,既保证了高维图像的生成质量,又提高了模型的收敛速度,并减少了参数量。实验结果表明,优化后的生成器结构将总运行时间提升了约92.88%,模型参数减少了约5.59%,并且在MNIST和FMNIST数据集上的收敛速度优于传统及混合量子经典模型,展示了其在高维图像生成中的潜力。

关键词: 图像生成, 量子计算, 参数化量子线路, 混合生成对抗网络, 矩阵乘积算符

Abstract: Neural networks play a pivotal role in artificial intelligence,particularly in image generation.As a popular algorithm in recent years,generative adversarial networks(GANs) have demonstrated superior performance in this area.Quantum computing,merging with traditional AI algorithms,accelerates processing speeds and enhances data security,making it especially suitable for managing high-dimensional data and optimization problems.Within this context,hybrid quantum-classical GANs show promising results.However,these models face challenges in generating high-dimensional images,and the inclusion of linear layers in generatorsresults in elevated parameter counts.Therefore,a hybrid quantum-classical compressed GAN model using matrix pro-duct operators is proposed.This model improves the structure of the block quantum generator,enabling the generation of multiple data blocks in a single call,which enhances efficiency.It integrates the nonlinear properties of classical networks with matrix product operators,ensuring high-quality image generation,speeding up model convergence,and reducing parameter counts.Expe-rimental results show that the optimized generator structure increases total runtime by approximately 92.88%,reduces model parameters by about 5.59%,and surpasses traditional and hybrid quantum-classical models in convergence speed on MNIST and FMNIST datasets,demonstrating its potential for high-dimensional image generation.

Key words: Image generation, Quantum computing, Parametric quantum circuits, Hybrid generative adversarial networks, Matrix product operators

中图分类号:

TP385

张曜麟, 刘晓楠, 杜帅岐, 廉德萌. 基于矩阵乘积算符的混合量子压缩经典生成对抗网络[J]. 计算机科学, 2025, 52(6): 74-81. https://doi.org/10.11896/jsjkx.240500017

ZHANG Yaolin, LIU Xiaonan, DU Shuaiqi, LIAN Demeng. Hybrid Quantum-classical Compressed Generative Adversarial Networks Based on Matrix Product Operators[J]. Computer Science, 2025, 52(6): 74-81. https://doi.org/10.11896/jsjkx.240500017

参考文献

[1]GOODFELLOW I J,POUGET-ABADIE J,MIRZA M B,et al.Generative Adversarial Nets[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems-Volume 2.2014:2672-2680.
[2]ZHU J Y,KRÄHENBÜHL P,SHECHTMAN E,et al.Generative Visual Manipulation on the Natural Image Manifold[C]//European Conference on Computer Vision.2016:597-613.
[3]LEDIG C,THEIS L,HUSZAR F,et al.Photo-realistic singleimage super-resolution using a generative adversarial network[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR).2017:4681-4690.
[4]ISOLA P,ZHU J Y,ZHOU T,et al.Image-to-image translation with conditional adversarial networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR).2017:1125-1134.
[5]GULRAJANI I,AHMED F,ARJOVSKY M,et al.Improvedtraining of wasserstein GANS[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:5769-5779.
[6]DENG C,YANG E,LIU T,et al.Unsupervised semantic-pre-serving adversarial hashing for image search[J].IEEE Transactions on Image Processing 28,2019,28(8):4032-4044.
[7]PRESKILL J.Quantum computing in the NISQ era and beyond[J].Quantum,2018,2(1):79.
[8]BIAMONTE J,WITTEK P,PANCOTTI N,et al.Quantum ma-chine learning[J].Nature,2017,549(7671):195-202.
[9]LLOYD S,WEEDBROOK C.Quantum generative adversariallearning[J].Physical Review Letters,2018,121(4):040502.
[10]ASSOUEL A,JACQUIER A,KONDRATYEV A.A quantumgenerative adversarial network for distributions[J].arXiv:2110.02742v1,2021.
[11]HUANG H L,DU Y,GONG M,et al.Experimental quantum generative adversarial networks for image generation[J].Physical Review Applied,2021,16(2):024051.
[12]CHU C,SKIPPER G,SWANY M,et al.IQGAN:robust quantum generative adversarial network for image synthesis on NISQ Devices[J].arXiv:2210.16857v1,2022.
[13]TSANG S L,WEST M T,ERFANI S M,et al.Hybrid quantum-classical generative adversarial network for high resolution image generation[J].arXiv:2212.11614v2,2023.
[14]JIANG Y D,WANG M M.Data reconstruction based on quantum generative adversarial networks[J].Computer Engineering and Applications,2024,60(5):156-164.
[15]CHEN S Y C,YANG C H H,QI J,et al.Variational quantum circuits for deep reinforcement learning[J].IEEE Access,2020,8(1):141007-141024.
[16]PIRVU B,MURG V,CIRAC J I,et al.Matrix product operator representations[J].New Journal of Physics,2010:12(2):025012.
[17]LIU X N,LIAN D M,DU S Q,et al.Simulation of limited entangled quantum fourier transform based on matrix product state[J].Computer Science,2024,51(9):80-86.
[18]GAO Z F,CHENG S,HE R Q,et al.Compressing deep neural networks by matrix product operators[J].Physical Review Research,2020,2(2):023300.
[19]LI J G,ZHAO H T,SUN S Y.KL-divergence-based policy optimization[J].Computer Science,2019,46(6):212-217.
[20]STEIN S A,BAHERI B,CHEN D,et al.QuGAN:a generative adversarial network through quantum states[J].arXiv:2010.09036,2020.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于矩阵乘积算符的混合量子压缩经典生成对抗网络

Hybrid Quantum-classical Compressed Generative Adversarial Networks Based on Matrix Product Operators

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

Metrics

本文评价

推荐阅读 0