Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.

A co-design framework of neural networks and quantum circuits towards quantum advantage

Open access
Jiang, Weiwen, Xiong, Jinjun, Shi, Yiyu · 2021 · Nature Communications

Published version of a preprint

Abstract

Despite the pursuit of quantum advantages in various applications, the power of quantum computers in executing neural network has mostly remained unknown, primarily due to a missing tool that effectively designs a neural network suitable for quantum circuit. Here, we present a neural network and quantum circuit co-design framework, namely QuantumFlow, to address the issue. In QuantumFlow, we represent data as unitary matrices to exploit quantum power by encoding n = 2 k inputs into k qubits and representing data as random variables to seamlessly connect layers without measurement. Coupled with a novel algorithm, the cost complexity of the unitary matrices-based neural computation can be reduced from O ( n ) in classical computing to O ( p o l y l o g ( n )) in quantum computing. Results show that on MNIST dataset, QuantumFlow can achieve an accuracy of 94.09% with a cost reduction of 10.85 × against the classical computer. All these results demonstrate the potential for QuantumFlow to achieve the quantum advantage.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-020-20729-5

Cites

53 of 53 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Illustration of the proposed end-to-end co-design framework, QuantumFlow. QuantumFlow provides a missing link between neural network and quantum circuit designs, which is composed of QF-pNet, QF-hNet, QF-FB, QF-Circ, QF-Map that work collaboratively to design neural networks and their quantum implementations.
Fig. 1 Illustration of the proposed end-to-end co-design framework, QuantumFlow. QuantumFlow provides a missing link between neural network and quantum circuit designs, which is composed of QF-pNet, QF-hNet, QF-FB, QF-Circ, QF-Map that work collaboratively to design neural networks and their quantum implementations.CC BY · PubMed Central
Fig. 2 Experimental results. a QF-hNet achieves state-of-the-art accuracy in image classifications on different sub-datasets of MNIST. b Output probability comparison on QF-FB(C), QF-F(Q)-ideal which assumes perfect qubits, QF-F(Q)-noise which applies noise model to “ibm_armonk” backend of IBM quantum processor, and results of circuit design (“design 4”) in subfigure (i) on the same backend. c Demonstration of Quantum Advantage achieved by U-LYR in QuantumFlow: comparison is conducted by using 50 random generated weights for each input size. d Design 1 for a two-input neural computation, which follows the original neural computation design. e–g Three optimized designs (design 2–4), based on design 1. h The deviation of design 1 and design 4 obtained from “ibm_velencia” backend IBM quantum processor, using QF-FB(C) as golden results.
Fig. 2 Experimental results. a QF-hNet achieves state-of-the-art accuracy in image classifications on different sub-datasets of MNIST. b Output probability comparison on QF-FB(C), QF-F(Q)-ideal which assumes perfect qubits, QF-F(Q)-noise which applies noise model to “ibm_armonk” backend of IBM quantum processor, and results of circuit design (“design 4”) in subfigure (i) on the same backend. c Demonstration of Quantum Advantage achieved by U-LYR in QuantumFlow: comparison is conducted by using 50 random generated weights for each input size. d Design 1 for a two-input neural computation, which follows the original neural computation design. e–g Three optimized designs (design 2–4), based on design 1. h The deviation of design 1 and design 4 obtained from “ibm_velencia” backend IBM quantum processor, using QF-FB(C) as golden results.CC BY · PubMed Central
Fig. 3 Results of a binary classification case study on IBM quantum processor of “ibmq_essex” backend. a Binary classification with two inputs “x” and “y”; b QF-Nets with trained parameters; c QF-Circ derived from the trained QF-Nets where input is encoded by f(x, y); d the virtual-to-physic mapping obtained by QF-Map upon “ibmq_essex” quantum processor; e QF-FB(C) achieves 100% accuracy; f QF-F(Q) achieves 98% accuracy where 2 marked error cases having probability deviation within 0.6%; g results on “ibmq_essex” using the default mapping, achieving 68% accuracy; h results obtained by “ibmq_essex” with the mapping in (d), achieving 82% accuracy; shots number in all tests is set as 8192.
Fig. 3 Results of a binary classification case study on IBM quantum processor of “ibmq_essex” backend. a Binary classification with two inputs “x” and “y”; b QF-Nets with trained parameters; c QF-Circ derived from the trained QF-Nets where input is encoded by f(x, y); d the virtual-to-physic mapping obtained by QF-Map upon “ibmq_essex” quantum processor; e QF-FB(C) achieves 100% accuracy; f QF-F(Q) achieves 98% accuracy where 2 marked error cases having probability deviation within 0.6%; g results on “ibmq_essex” using the default mapping, achieving 68% accuracy; h results obtained by “ibmq_essex” with the mapping in (d), achieving 82% accuracy; shots number in all tests is set as 8192.CC BY · PubMed Central
Fig. 4 Neural computation. a Prepossessing of inputs by (i) downsampling the original 28 × 28 image in MNIST to 4 × 4 image and (ii) get the 4 × 4 matrix and normalize gray level to [0, 1]. b In P-LYR, input data are converted from real number to a random variable following a two-point distribution. c Four operations in P-LYR, (i) R: converting a real number ranging from 0 to 1 to a random variable, (ii) C: average sum of weighted inputs, (iii) A: nonlinear activation function, (iv) E: converting random variable to a real number. d In U-LYR, m-input data are converted to a vector in the first column of a m × m unitary matrix. e Three operations in U-LYR, (i) U: unitary matrix converter, (ii) Cu: average sum of weighted inputs; (iii) Au: nonlinear activation function. f Normalization for the connection of P-LYR. g Normalization for the connection of U-LYR.
Fig. 4 Neural computation. a Prepossessing of inputs by (i) downsampling the original 28 × 28 image in MNIST to 4 × 4 image and (ii) get the 4 × 4 matrix and normalize gray level to [0, 1]. b In P-LYR, input data are converted from real number to a random variable following a two-point distribution. c Four operations in P-LYR, (i) R: converting a real number ranging from 0 to 1 to a random variable, (ii) C: average sum of weighted inputs, (iii) A: nonlinear activation function, (iv) E: converting random variable to a real number. d In U-LYR, m-input data are converted to a vector in the first column of a m × m unitary matrix. e Three operations in U-LYR, (i) U: unitary matrix converter, (ii) Cu: average sum of weighted inputs; (iii) Au: nonlinear activation function. f Normalization for the connection of P-LYR. g Normalization for the connection of U-LYR.CC BY · PubMed Central
Fig. 5 QF-FB: the determination of parameters, t, θ, and γ, in batch normalization component of QF-Nets. a Batch_adj when t = 1; b batch_adj when t = 0; c indiv_adj.
Fig. 5 QF-FB: the determination of parameters, t, θ, and γ, in batch normalization component of QF-Nets. a Batch_adj when t = 1; b batch_adj when t = 0; c indiv_adj.CC BY · PubMed Central
Fig. 6 The corresponding implementation of operations (e.g., R, C, A) in QF-Circ. a Quantum circuit designs for QF-pNet; b quantum circuit design for QF-hNet; c–f batch normalization, quantum circuit designs for different cases; c design of “batch_adj” for the case of t = 0; d design of “batch_adj” for the case of t = 1; e design of “indiv_adj”; f optimized design for a specific case when t = 1 in “batch_adj”.
Fig. 6 The corresponding implementation of operations (e.g., R, C, A) in QF-Circ. a Quantum circuit designs for QF-pNet; b quantum circuit design for QF-hNet; c–f batch normalization, quantum circuit designs for different cases; c design of “batch_adj” for the case of t = 0; d design of “batch_adj” for the case of t = 1; e design of “indiv_adj”; f optimized design for a specific case when t = 1 in “batch_adj”.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 51 references

Research works printed but not linked: 6

  1. Goodfellow, I., Bengio, Y. & Courville, A.Deep Learning (MIT Press, 2016).
  2. Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition. International Conference on Learning Representations. (2015).
  3. Bonaldi, A. & Braun, R. Square kilometre array science data challenge 1. Preprint at https://arxiv.org/abs/1811.10454 (2018). · arXiv
  4. Nielsen, M. A., & Chuang, I. L. Quantum Computation and Quantum Information. (Cambridge University Press, 2010).
  5. Di Matteo, O., Gheorghiu, V. & Mosca, M. Fault-tolerant resource estimation of quantum random-access memories. IEEE Trans. Quantum Eng. 1, 1–13 (2020). · DOI
  6. Bausch, J. Fast black-box quantum state preparation. Preprint at https://arxiv.org/abs/2009.10709 (2020). · arXiv

Cited by

See all 77

Metadata

Authors
Jiang, W., Xiong, J., & Shi, Y.
Affiliations
  • University of Notre Dame (US)
  • IBM (United States) (US)
Corresponding author
  • Jiang, Weiwen
  • Xiong, Jinjun
  • Shi, Yiyu
Venue
Nature Communications
Published
2021-01-25
Volume
12
Issue
1
Pages
579-579
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3084318570
Cited by
63
References
51
Citations by year
2026: 6 · 2025: 10 · 2024: 16 · 2023: 9 · 2022: 14 · 2021: 7
FWCI
4.06
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • Research Square (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • arXiv (Cornell University)
Funders
  • University of Notre Dame
  • Center for Cognitive Computing Systems Research
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum many-body systems
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum Information and Cryptography
Keywords
quantum neural networks, quantum circuits, co-design, quantum advantage, quantum machine learning, MNIST classification

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: A co-design framework of neural networks and quantum circuits towards quantum advantage. https://app.cochat.ai/papers/10.1038/s41467-020-20729-5
The paper
Jiang, W., Xiong, J., & Shi, Y. (2021). A co-design framework of neural networks and quantum circuits towards quantum advantage. Nature Communications. https://doi.org/10.1038/s41467-020-20729-5

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.