pith. sign in

arxiv: 2606.07657 · v1 · pith:QHKI5NJJnew · submitted 2026-06-03 · 💻 cs.NE · cs.LG

QDS-SNN: Energy-efficient Quantum Deeply-Supervised Spiking Neural Network Algorithm for Traffic Sign Recognition

Zhiguo Qu , Keqi Li , Le Sun , Wenjie Liu , Yimin Yu , Saif Al-Kuwari , Ahmed Farouk This is my paper

Pith reviewed 2026-06-28 03:31 UTC · model grok-4.3

classification 💻 cs.NE cs.LG
keywords spiking neural networksquantum neural networkstraffic sign recognitionenergy efficiencydeep supervisionGTSRB datasetautonomous drivingLIF neurons
0
0 comments X

The pith

A quantum-deeply-supervised spiking neural network achieves 99.72% accuracy on traffic sign recognition in only 6 time steps while reducing energy consumption by over 55%.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces QDS-SNN to address limitations in spiking neural networks for traffic sign recognition, such as information loss and vanishing gradients. It integrates quantum neural networks to provide efficient deep supervision using superposition and entanglement for better representations. The model includes a TSA-LIF neuron and QACM to improve training. On the GTSRB dataset, it reaches 99.72% accuracy in 6 steps, beating the MS-ResNet baseline by 1.32% with 55.77% less energy. Similar gains appear on the TSRD dataset, showing a path to high-performance, low-power recognition for intelligent transportation.

Core claim

QDS-SNN combines quantum neural networks with spiking neural networks via deep supervision, TSA-LIF neurons, and a quantum-assisted classifier module, yielding 99.72% accuracy on GTSRB and 97.90% on TSRD while cutting energy use to roughly half of the baseline.

What carries the argument

The quantum-assisted classifier module (QACM) combined with the temporally and spatially adaptive LIF (TSA-LIF) neuron, which uses quantum superposition for parallel computation and adaptive thresholds to reduce gradient problems.

If this is right

  • QDS-SNN enables real-time traffic sign recognition suitable for autonomous driving systems with lower power requirements.
  • The approach demonstrates that quantum elements can enhance SNN performance without sacrificing their energy advantages.
  • Deep supervision in SNNs becomes feasible through quantum integration, mitigating vanishing gradient issues.
  • Energy reductions of over 50% make deployment on resource-constrained devices more practical.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Similar quantum-SNN hybrids could apply to other spatiotemporal tasks like video processing or sensor data analysis.
  • Further scaling might allow training on smaller datasets due to the expressive power of quantum representations.
  • Integration with actual quantum hardware could amplify the efficiency gains beyond simulation results.

Load-bearing premise

The reported accuracy gains and energy savings stem directly from the quantum deep supervision and the TSA-LIF/QACM components rather than from differences in training procedures or baseline implementations.

What would settle it

An experiment that applies the same training setup and architecture without the quantum modules and measures whether accuracy drops below 98.4% or energy use rises above the baseline on the GTSRB dataset.

Figures

Figures reproduced from arXiv: 2606.07657 by Ahmed Farouk, Keqi Li, Le Sun, Saif Al-Kuwari, Wenjie Liu, Yimin Yu, Zhiguo Qu.

Figure 1
Figure 1. Figure 1: The Proposed QDS-SNN Framework. ule (QACM) and the Spatio-Temporal Adaptive LIF neuron (TSALIF) into a K-stage deep SNN framework. Deep SNNs commonly suffer from vanishing gradients due to long temporal dependencies and non-differentiable activation functions. To address this issue, QACM is added after each stage to enhance gradient propagation through quantum convolution, pooling, and measurement. Traditi… view at source ↗
Figure 2
Figure 2. Figure 2: Temporal-Spatial Adaptive-LIF (TSA-LIF). [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Quantum-Assisted Classification Module (QACM). [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Quantum Convolution Layer Circuit. = = 4 qubits Pool 2 qubits Pool [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Quantum Pooling Layer Circuit. reducing gate depth and resource use while preserving quan￾tum expressiveness. Quantum Pooling Layer: To compress the quantum state’s dimensions, extract global semantics, and improve general￾ization, this study introduces a quantum pooling layer after the quantum convolution layer ( [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Traffic sign recognition dataset example. (a) GTSRB dataset example [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Ablation study of deep supervision factor [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Convergence comparison of QDS-SNN on different time steps on the [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Convergence comparison of QDS-SNN on different time steps on the [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Fidelity of QNN under different quantum noise when p=0.01(left) [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
read the original abstract

Traffic sign recognition is crucial for intelligent transportation and autonomous driving, as it can improve driving efficiency and ensure road safety. However, traditional recognition methods are based on large datasets and intensive computation, which limits their real-time applicability. Spiking Neural Networks (SNNs) offer a biologically inspired, energy-efficient alternative due to their spatiotemporal processing capabilities, but suffer from information loss and vanishing gradients during training. To overcome these limitations, this study proposes a Quantum Deep-supervised Spiking Neural Network (QDS-SNN) that integrates Quantum Neural Networks (QNNs) for efficient, low-power deep supervision. Using quantum superposition and entanglement, QNNs enable expressive representations and parallel computation, thereby enhancing performance without compromising energy efficiency. The proposed QDS-SNN incorporates a temporally and spatially adaptive LIF (TSA-LIF) neuron and a quantum-assisted classifier module (QACM) to mitigate gradient issues and improve training effectiveness. This study conducts experiments on the PennyLane quantum simulation platform, and the results show that QDS-SNN achieves 99.72\% accuracy on the GTSRB dataset in only 6 time steps -- outperforming the MS-ResNet baseline by 1.32\% while reducing energy consumption by 55.77\%. In the TSRD dataset, it achieves 97.90\% accuracy while reducing energy use to 52.68\% of the baseline. These results demonstrate that QDS-SNN offers a high-performance, energy-efficient solution for traffic sign recognition in intelligent transportation systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The manuscript proposes QDS-SNN, a spiking neural network augmented with quantum neural networks for deep supervision. It introduces a temporally and spatially adaptive LIF (TSA-LIF) neuron and a quantum-assisted classifier module (QACM), and reports results from PennyLane simulations on the GTSRB and TSRD traffic-sign datasets: 99.72% accuracy on GTSRB in 6 time steps (1.32% above an MS-ResNet baseline) together with a 55.77% energy reduction, and 97.90% accuracy on TSRD with energy consumption at 52.68% of the baseline.

Significance. If the reported accuracy and energy gains can be shown to arise specifically from the quantum deep-supervision mechanism rather than from unstated implementation choices, the work would offer a concrete example of hybrid quantum-classical neuromorphic models for real-time perception tasks. The explicit use of a quantum simulator for SNN training and the short 6-step inference latency are potentially useful contributions, but the absence of supporting controls prevents assessment of whether these constitute a genuine advance over existing SNN or QNN-SNN hybrids.

major comments (3)
  1. [Abstract and Results section] Abstract and Results section: the central claims of 1.32% accuracy improvement and 55.77% energy reduction on GTSRB (and the analogous TSRD figures) are stated without any ablation tables, non-quantum deep-supervision controls, or matched-parameter baseline re-runs that would isolate the contribution of the QNN integration, TSA-LIF adaptation rates, and QACM from differences in total parameters, spike encoding, or training schedule.
  2. [Methods/Experimental setup (implied)] Methods/Experimental setup (implied): no explicit energy model is supplied (gate count, circuit depth, or simulator proxy) and no verification is given that quantum-circuit overhead is included in the reported energy figures, rendering the 55.77% and 52.68% reductions impossible to interpret or reproduce.
  3. [Results section] Results section: the comparison to the MS-ResNet baseline provides no information on whether the baseline was re-trained under identical conditions or whether parameter counts and spike-encoding schemes were matched, which is required to support the attribution of gains to the proposed quantum-assisted components.
minor comments (1)
  1. [Abstract] Abstract: the phrases 'quantum superposition and entanglement' are invoked without a concrete description of how these properties are realized inside the QACM or the deep-supervision pathway.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight areas where additional controls and documentation will strengthen the manuscript. We address each point below and will incorporate the requested clarifications and analyses in the revision.

read point-by-point responses

  1. Referee: [Abstract and Results section] Abstract and Results section: the central claims of 1.32% accuracy improvement and 55.77% energy reduction on GTSRB (and the analogous TSRD figures) are stated without any ablation tables, non-quantum deep-supervision controls, or matched-parameter baseline re-runs that would isolate the contribution of the QNN integration, TSA-LIF adaptation rates, and QACM from differences in total parameters, spike encoding, or training schedule.

    Authors: We agree that ablation studies are needed to isolate the contributions of the QNN integration, TSA-LIF neuron, and QACM. In the revised manuscript we will add ablation tables that systematically remove or replace each component while keeping total parameter count, spike encoding, and training schedule fixed. We will also include a non-quantum deep-supervision control (standard deep supervision without QACM) to better attribute the reported gains. revision: yes

  2. Referee: [Methods/Experimental setup (implied)] Methods/Experimental setup (implied): no explicit energy model is supplied (gate count, circuit depth, or simulator proxy) and no verification is given that quantum-circuit overhead is included in the reported energy figures, rendering the 55.77% and 52.68% reductions impossible to interpret or reproduce.

    Authors: We acknowledge the absence of an explicit energy model. The revised Methods section will provide the precise energy estimation procedure used with PennyLane, including the gate-count and circuit-depth proxy, the mapping from quantum operations to energy cost, and explicit confirmation that QACM overhead is folded into the reported figures. This will enable reproduction of the energy numbers. revision: yes

  3. Referee: [Results section] Results section: the comparison to the MS-ResNet baseline provides no information on whether the baseline was re-trained under identical conditions or whether parameter counts and spike-encoding schemes were matched, which is required to support the attribution of gains to the proposed quantum-assisted components.

    Authors: We will expand the experimental-setup subsection to state that the MS-ResNet baseline was re-trained from scratch under the identical training schedule, optimizer, and data augmentation used for QDS-SNN. We will also report the exact parameter counts and confirm that the same spike-encoding scheme was applied to both models, thereby supporting direct attribution of the observed differences. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical claims rest on reported experiments

full rationale

The manuscript proposes QDS-SNN integrating QNNs, TSA-LIF neurons and QACM, then reports measured accuracy (99.72 % on GTSRB, 97.90 % on TSRD) and relative energy figures against an MS-ResNet baseline. No derivation chain, equations, or first-principles steps are supplied in the abstract or visible claims; performance numbers are presented as simulation outcomes on PennyLane rather than quantities obtained by algebraic reduction to the model definition itself. No self-citations, fitted-input renamings, or ansatz smuggling appear. The result is therefore self-contained as an empirical architecture-plus-experiment report.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 2 invented entities

Abstract-only review supplies insufficient detail to enumerate free parameters, axioms, or invented entities with precision; listed items are inferred directly from stated components.

free parameters (1)
  • TSA-LIF adaptation rates
    Adaptive parameters in the neuron model are introduced without stated values or derivation, implying they are chosen or fitted.
axioms (1)
  • domain assumption Quantum superposition and entanglement enable expressive representations and parallel computation in neural networks
    Invoked in the abstract as the basis for QNN benefits.
invented entities (2)
  • TSA-LIF neuron no independent evidence
    purpose: Mitigate gradient issues and improve training effectiveness in SNNs
    New neuron variant introduced to address limitations
  • QACM no independent evidence
    purpose: Quantum-assisted classifier module for improved classification
    New module added to the network architecture

pith-pipeline@v0.9.1-grok · 5831 in / 1344 out tokens · 55198 ms · 2026-06-28T03:31:03.075363+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

46 extracted references · 4 canonical work pages · 1 internal anchor

  1. [2]

    Z. Qu, Y . Xia, L. Sun, W. Liu, and G. Muhammad, ”QCACNN: A quan- tum convolutional neural network algorithm for traffic sign recognition in carbon-intelligent electric vehicles,”IEEE Internet Things J., vol. 12, no. 17, pp. 34708–34719, 2025

    2025

  2. [3]

    M. Q. Kheder and A. A. Mohammed, ”Improved traffic sign recognition system (ITSRS) for autonomous vehicle based on deep convolutional neural network,”Multimed. Tools Appl., vol. 83, no. 22, pp. 61821–61841, 2024

    2024

  3. [4]

    Shustanov and P

    A. Shustanov and P. Yakimov, ”CNN design for real-time traffic sign recognition,”Procedia Eng., vol. 201, pp. 718–725, 2017

    2017

  4. [5]

    J. Hu, L. Shen, and G. Sun, ”Squeeze-and-excitation networks,” inProc. IEEE Conf. Comput. Vis. Pattern Recognit., pp. 7132–7141, 2018

    2018

  5. [6]

    Hasan, T

    N. Hasan, T. Anzum, and N. Jahan, ”Traffic sign recognition system (TSRS): SVM and convolutional neural network,” inInvent. Commun. Comput. Technol.: Proc. ICICCT 2020, pp. 69–79, 2020

    2020

  6. [7]

    Traffic sign recognition using optimized federated learning in internet of vehicles,

    Z. Lian, Q. Zeng, W. Wang, D. Xu, W. Meng, and C. Su, “Traffic sign recognition using optimized federated learning in internet of vehicles,” IEEE Internet Things J., vol. 11, no. 4, pp. 6722–6729, 2023

    2023

  7. [8]

    SLMFed: A stage-based and layerwise mechanism for incremental federated learning to assist dynamic and ubiquitous IoT,

    L. You, Z. Guo, B. Zuo, Y . Chang, and C. Yuen, “SLMFed: A stage-based and layerwise mechanism for incremental federated learning to assist dynamic and ubiquitous IoT,”IEEE Internet Things J., vol. 11, no. 9, pp. 16364–16381, 2024

    2024

  8. [9]

    K. Xie, Z. Zhang, Bo. Li, et al., ”Efficient federated learning with spike neural networks for traffic sign recognition,”IEEE Trans. Veh. Technol., vol. 71, no. 9, pp. 9980–9992, 2022

    2022

  9. [10]

    Zhang, H

    Y . Zhang, H. Xu, L. Huang, and C. Chen, ”A storage-efficient SNN– CNN hybrid network with RRAM-implemented weights for traffic signs recognition,”Eng. Appl. Artif. Intell., vol. 123, p. 106232, 2023. INTERNET OF THINGS JOURNAL, VOL. , NO. , 13

    2023

  10. [11]

    Traffic sign recognition model based on spiking neural network,

    H. Chen, Y . Liu, W. Ye, and C. Han, “Traffic sign recognition model based on spiking neural network,” in2024 4th Int. Conf. Neural Netw., Inf. Commun. Eng. (NNICE 2024), pp. 693–696, 2024

    2024

  11. [12]

    Yadav and B

    C. Yadav and B. S. Reniwal, ”Highly Accurate and Energy Efficient Convolutional Spiking Neural Network for Traffic Sign Recognition,” IEEE Access, vol. 13, pp. 212341–212354, 2025

    2025

  12. [13]

    Zhang, K

    T. Zhang, K. Yu, J. Zhang, and H. Wang, ”Da-LIF: Dual adaptive leaky integrate-and-fire model for deep spiking neural networks,” in2025 IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP), pp. 1–5, 2025

    2025

  13. [14]

    G. Li, L. Deng, H. Tang, et al., ”Brain-inspired computing: A systematic survey and future trends,”Proc. IEEE, vol. 112, no. 6, pp. 544–584, 2024

    2024

  14. [15]

    Y . Li, S. Deng, X. Dong, R. Gong, and S. Gu, ”A free lunch from ANN: Towards efficient, accurate spiking neural networks calibration,” inProc. Int. Conf. Mach. Learn. (ICML), pp. 6316–6325, 2021

    2021

  15. [16]

    Kaiser, J

    J. Kaiser, J. C. V . Tieck, C. Hubschneider, et al., ”Towards a frame- work for end-to-end control of a simulated vehicle with spiking neural networks,” inProc. 2016 IEEE Int. Conf. Simul. Model. Program. Auton. Robots (SIMPAR), pp. 127–134, 2016

    2016

  16. [17]

    Z. Bing, C. Meschede, G. Chen, A. Knoll, and K. Huang, ”Indirect and direct training of spiking neural networks for end-to-end control of a lane-keeping vehicle,”Neural Netw., vol. 121, pp. 21–36, 2020

    2020

  17. [18]

    Chandarana, J

    P. Chandarana, J. Ou, and R. Zand, ”An adaptive sampling and edge de- tection approach for encoding static images for spiking neural networks,” inProc. 2021 12th Int. Green Sustain. Comput. Conf. (IGSC), pp. 1–8, 2021

    2021

  18. [19]

    Kim and P

    Y . Kim and P. Panda, ”Revisiting batch normalization for training low- latency deep spiking neural networks from scratch,”Front. Neurosci., vol. 15, p. 773954, 2021

    2021

  19. [20]

    Zheng, Y

    H. Zheng, Y . Wu, L. Deng, Y . Hu, and G. Li, ”Going deeper with directly-trained larger spiking neural networks,” inProc. AAAI Conf. Artif. Intell., vol. 35, no. 12, pp. 11062–11070, 2021

    2021

  20. [21]

    S. Deng, Y . Li, S. Zhang, and S. Gu, ”Temporal efficient training of spiking neural network via gradient re-weighting,”arXiv:2202.11946, 2022

    work page arXiv 2022

  21. [22]

    H. Chen, Y . Liu, W. Ye, and C. Han, ”Traffic sign recognition model based on spiking neural network,” inProc. 2024 4th Int. Conf. Neural Netw. Inf. Commun. Eng. (NNICE), pp. 693–696, 2024

    2024

  22. [23]

    Z. Zhou, Y . Zhu, C. He, Y . Wang, S. Yan, Y . Tian, and L. Yuan, ”Spikformer: When spiking neural network meets transformer,” arXiv:2209.15425, 2022

    work page arXiv 2022

  23. [24]

    Y . Hu, L. Deng, Y . Wu, M. Yao, and G. Li, ”Advancing spiking neural networks toward deep residual learning,”IEEE Trans. Neural Netw. Learn. Syst., vol. 36, no. 2, pp. 2353–2367, 2024

    2024

  24. [25]

    Y . Ding, L. Zuo, M. Jing, P. He, and Y . Xiao, ”Shrinking your timestep: Towards low-latency neuromorphic object recognition with spiking neural networks,” inProc. AAAI Conf. Artif. Intell., vol. 38, no. 10, pp. 11811–11819, 2024

    2024

  25. [26]

    K. Yu, T. Zhang, Q. Xu, G. Pan, and H. Wang, ”TS-SNN: Temporal shift module for spiking neural networks,”arXiv:2505.04165, 2025

    work page arXiv 2025

  26. [27]

    Highly Accurate and Energy Efficient Convolutional Spiking Neural Network for Traffic Sign Recognition,

    C. Yadav, and B. S. Reniwal, “Highly Accurate and Energy Efficient Convolutional Spiking Neural Network for Traffic Sign Recognition,” IEEE Access, vol. 13, pp. 212341–212354, 2025

    2025

  27. [28]

    S. Shen, J. Yang, H. Zhong, H. Lu, X. Zheng, and H. Yang, ”Deployment-friendly lane-changing intention prediction powered by brain-inspired spiking neural networks,” in2025 IEEE Intell. Veh. Symp. (IV), pp. 2310–2316, 2025

    2025

  28. [29]

    S. C. Kak, ”Quantum neural computing,”Adv. Imaging Electron Phys., vol. 94, pp. 259–313, 1995

    1995

  29. [30]

    Z. Hao, T. Bu, J. Ding, T. Huang, and Z. Yu, ”Reducing ANN–SNN conversion error through residual membrane potential,” inProc. AAAI Conf. Artif. Intell., vol. 37, no. 1, pp. 11–21, 2023

    2023

  30. [31]

    W. Fang, Z. Yu, Y . Chen, T. Huang, T. Masquelier, and Y . Tian, ”Deep residual learning in spiking neural networks,” inAdv. Neural Inf. Process. Syst., vol. 34, pp. 21056–21069, 2021

    2021

  31. [32]

    Wiebe, A

    N. Wiebe, A. Kapoor, and K. M. Svore, ”Quantum algorithms for nearest-neighbor methods for supervised and unsupervised learning,” Quan. Inf. Com., vol. 15, nos. 3–4, pp. 316–356, 2015

    2015

  32. [33]

    Preskill, ”Quantum computing in the NISQ era and beyond,”Quan- tum, vol

    J. Preskill, ”Quantum computing in the NISQ era and beyond,”Quan- tum, vol. 2, p. 79, 2018

    2018

  33. [34]

    Classification with Quantum Neural Networks on Near Term Processors

    E. Farhi and H. Neven, ”Classification with quantum neural networks on near-term processors,”arXiv:1802.06002, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  34. [35]

    Schuld, A

    M. Schuld, A. Bocharov, K. M. Svore, and N. Wiebe, ”Circuit-centric quantum classifiers,”Phys. Rev. A, vol. 101, no. 3, p. 032308, 2020

    2020

  35. [36]

    Z. Qu, Z. Chen, S. Dehdashti, et al., ”QFSM: A novel quantum federated learning algorithm for speech emotion recognition with minimal gated unit in 5G IoV ,”IEEE Trans. Intell. Veh., vol. 9, no. 10, pp. 6512–6523, 2024

    2024

  36. [37]

    DAQFL: Dynamic aggregation quantum federated learning algorithm for intelligent diagnosis in Internet of Medical Things,

    Z. Qu, X. Zhao, L. Sun, and G. Muhammad, “DAQFL: Dynamic aggregation quantum federated learning algorithm for intelligent diagnosis in Internet of Medical Things,”IEEE Internet Things J., vol. 12, no. 19, pp. 39313–39325, 2025

    2025

  37. [38]

    Brand and F

    D. Brand and F. Petruccione, ”A quantum leaky integrate-and-fire spiking neuron and network,”npj Quant. Inf., vol. 10, no. 1, pp. 1–8, 2024

    2024

  38. [39]

    A hybrid spiking neural network-quantum framework for spatio-temporal data classification: a case study on EEG data,

    R. K. Jha, N. Kasabov, S. Bhattacharyya, et al., “A hybrid spiking neural network-quantum framework for spatio-temporal data classification: a case study on EEG data,”EPJ Quantum Technol., vol. 12, no. 1, pp. 1–23, 2025

    2025

  39. [40]

    Liu and Y

    S. Liu and Y . Gu, ”Quantum spiking neural networks for image classification,” inProc. 3th Int. Conf. Alg., Network Com. Tech.(ICANCT), vol. 13545, pp. 181–188, 2025

    2025

  40. [41]

    Y . Wu, L. Deng, G. Li, J. Zhu, and L. Shi, ”Spatio-temporal backprop- agation for training high-performance spiking neural networks,”Front. Neurosci., vol. 12, p. 331, 2018

    2018

  41. [42]

    Schuld, V

    M. Schuld, V . Bergholm, C. Gogolin, J. Izaac, and N. Killoran, ”Eval- uating analytic gradients on quantum hardware,”Phys. Rev. A, vol. 99, no. 3, p. 032331, 2019

    2019

  42. [43]

    Stallkamp, M

    J. Stallkamp, M. Schlipsing, J. Salmen, and C. Igel, ”The German traffic sign recognition benchmark: A multi-class classification competition,” in Proc. Int. Joint Conf. Neural Netw., pp. 1453–1460, 2011

    2011

  43. [44]

    National Natural Science Foundation of China, ”Chinese traffic sign database,”[Online].https://nlpr.ia.ac.cn/pal/trafficdata/recognition.html

  44. [45]

    M. Yao, J. Hu, Z. Zhou, L. Yuan, Y . Tian, B. Xu, and G. Li, ”Spike- driven transformer,” inAdv. Neural Inf. Process. Syst., vol. 36, pp. 64043–64058, 2023

    2023

  45. [46]

    Killoran, J

    N. Killoran, J. Izaac, N. Quesada, et al., ”Strawberry fields: A software platform for photonic quantum computing,”Quantum, vol. 3, p. 129, 2019

    2019

  46. [47]

    M. A. Nielsen and I. L. Chuang, ”Quantum computation and quantum information,”Quantum Inf. Comput., 2010. Zhiguo Quis currently a Professor in the School of Com- puter Science. His research interests include quantum machine learning, quantum communications, and quantum computing. Keqi Liis currently a master student at Nanjing University of Information Sc...

    2010