Quantum Machine Learning for UAV Swarm Intrusion Detection

Chen-Yu Liu; Kin K. Leung; Kuan-Cheng Chen; Samuel Yen-Chi Chen; Tai-Yue Li

REVIEW 4 major objections 6 minor 38 references

Quantum Machine Learning for UAV Swarm Intrusion Detection

T0 review · 4 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A hybrid quantum-classical network wins UAV swarm intrusion detection on the paper's simulated benchmark.

desk verdict Plausible hybrid QNN benchmark on UAVIDS-2025, but the abstract promises a CNN comparison the experiments never make. read the letter →

arxiv 2509.01812 v1 pith:JYUM5PFV submitted 2025-09-01 quant-ph cs.AIcs.SYeess.SY

Kuan-Cheng Chen , Samuel Yen-Chi Chen , Tai-Yue Li , Chen-Yu Liu , Kin K. Leung This is my paper

classification quant-phcs.AIcs.SYeess.SY

keywords quantummachinelearningintrusiondetectionUAVswarmshybridneuralnetworkskernelsclassimbalancenetworksecurityvariationalcircuits

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper asks whether quantum machine learning can improve intrusion detection for UAV swarms, where traffic is highly mobile, non-stationary, and class-imbalanced. It benchmarks three quantum approaches—quantum kernels, variational quantum neural networks, and quantum-trained neural networks—against classical baselines on a 120k-flow simulated UAV network corpus. Its central claim is that a shallow eight-qubit Hybrid QNN, combining a quantum encoder with a small classical output head, outperforms all other quantum and classical models under identical preprocessing and balancing conditions. The authors read this as evidence that hybrid quantum-classical designs, not pure variational circuits, are the most practical near-term route to quantum advantage in network-security workloads.

What carries the argument

The central object is the Hybrid QNN architecture: a hardware-efficient, shallow quantum circuit that encodes eight flow features into an eight-qubit state and produces measurement observables, followed by a small classical post-processing layer (18 parameters) that calibrates the outputs. This division of labor—quantum circuit for correlation extraction, classical head for normalization and gradient-noise mitigation—is what lets the model avoid barren-plateau regions and balance sensitivity against specificity, giving it the edge over pure QNNs, quantum kernels, and QT-NNs.

What would settle it

Run the same eight-layer Hybrid QNN and the classical SVM/CNN baselines on real UAV swarm traffic from a physical testbed, or on a second independently generated UAV network-flow corpus; if the hybrid model no longer matches or beats the classical baselines on accuracy and F1, the paper's central ranking claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that an eight-layer Hybrid QNN—an eight-qubit parameterized circuit whose measurement statistics are refined by a small classical head—achieves the best intrusion-detection performance on the UAVIDS-2025 corpus, reaching 0.948 accuracy, 0.967 F1, 0.972 sensitivity, and 0.838 specificity with only 18 classical and 64 quantum trainable parameters. This result is contrasted with three failure or partial-success patterns: pure deep QNNs suffer trainability problems (specificity collapses to as low as 0.04 while sensitivity approaches 0.99), quantum-kernel SVMs inherit the classical kernel's expressivity ceiling, and QT-NNs and Hybrid QNNs rebalance the sensitivity-sp

Load-bearing premise

The paper's performance ranking depends on the simulated UAV network corpus standing in for real UAV swarm traffic; if simulated flows do not reflect operational conditions, the model comparisons and conclusions may not transfer.

Editorial extensions

If this is right

Hybrid quantum-classical designs, rather than pure variational circuits, become the recommended template for near-term QML in network security workloads.
Pure deep QNNs are currently bottlenecked by trainability and noise: they become extremely sensitive but flood the system with false positives, so deployment-focused work should avoid them without mitigation.
Quantum kernels and QT-NNs provide value in low-data, nonlinear regimes, while classical CNNs remain competitive when abundant data offset their larger parameter counts.
An eight-qubit, 64-quantum-parameter model is sufficient to match or beat classical baselines on this benchmark, suggesting quantum resource footprints can stay small for flow-level IDS.
The specificity collapse observed in deeper QNNs offers a concrete diagnostic: high sensitivity with near-zero specificity indicates over-activation and trainability failure, not a useful detector.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Editorial: The model ranking rests entirely on a simulated corpus; the natural next test is running the same Hybrid QNN and baselines on real UAV swarm traffic or a physical testbed, where channel effects and hardware noise may change the ordering.
Editorial: The eight-feature flow representation is domain-agnostic enough that the hybrid architecture could be transferred to other flow-based intrusion-detection datasets (e.g., conventional network IDS corpora) with minimal changes, making the reported advantage testable outside UAV settings.
Editorial: The QT-NN parameter-compression argument suggests a broader test: stress the quantum weight-generation scheme on much larger classical networks to see whether the O(poly(log M)) trainable-parameter footprint persists without accuracy loss.
Editorial: The sensitivity-specificity trade-off pattern implies a simple heuristic for practitioners: if a variational model's specificity collapses below roughly 0.1 while sensitivity saturates, suspect trainability/over-activation rather than a genuinely high-performing detector.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear X,

The headline result—Hybrid QNN-8L at 0.948 accuracy, 0.967 F1—is a plausible, honestly reported benchmark on the UAVIDS-2025 dataset, and the comparison of pure QNN, quantum kernel, and hybrid designs under one feature set is genuinely new. But the abstract and conclusion promise a classical CNN baseline that never materializes in the experiments. Table I's only classical entry is an SVM. So the phrase 'outperforms all other quantum and classical contenders' is not backed by the data shown. That is the one thing you should know before citing it.

What the paper does well: it takes a fresh public dataset that fits the QML resource story, defines an interpretable 8-feature representation, and evaluates a range of quantum models—kernel, variational, hybrid, quantum-train—under the same preprocessing and balancing. The resource table (qubits, parameters) is useful, and the conclusion that deep pure QNNs collapse into trivial specificity (0.04–0.23) while hybrid readouts restore balance is a clean, credible empirical observation. The QT-NN compression idea from the authors' prior work is applied here, and the self-citation is legitimate: the cuTN-QSVM simulator is public, and the benchmark itself is external to it.

Soft spots, in order of severity. First, the missing CNN. The abstract claims 'CNNs dominate when abundant data offset their larger parameter count,' and the conclusion repeats the 'all classical contenders' line, but no CNN or any other classical neural net appears in the paper. That is an internal inconsistency—a stated result that is never run. If a well-tuned CNN or MLP reaches 0.95 accuracy on the same balanced subset, the central ranking collapses. Second, every metric is a single-run point estimate. No standard deviation, no repeated seeds, so you cannot tell whether the 0.948 vs 0.944 gap between Hybrid-8L and Hybrid-10L means anything. Third, the data availability section says the complete codebase is publicly released, but no link appears (only the cuTN-QSVM link). That claim is currently false as written. Fourth, the quantum-kernel low-data advantage is asserted in the abstract but not tested—no learning-curve experiment is shown.

None of these are fatal to the empirical benchmark itself; they are missing experiments and an overreaching abstract. As it stands, the paper is a useful data point but not the definitive comparison it claims to be. I would send it to peer review, with the expectation that the authors add the CNN baseline, repeated seeds, and fix the code link.

Best, [Your name]

Referee Report

4 major / 6 minor

Summary. The paper benchmarks three families of quantum machine learning models—quantum kernel methods, variational quantum neural networks (QNNs), and quantum-trained neural networks (QT-NNs)—against a classical SVM baseline on the UAVIDS-2025 simulated UAV swarm intrusion detection dataset. All models use an 8-feature flow representation and a balanced subset, with metrics including accuracy, macro-F1, specificity, sensitivity, and quantum resource counts. The headline result is that an eight-layer Hybrid QNN achieves the best accuracy (0.948), F1 (0.967), sensitivity (0.972), and specificity (0.838) among the models in Table I. The abstract additionally claims that quantum kernels and QT-NNs excel in low-data, nonlinear regimes and that CNNs dominate when data are abundant.

Significance. If fully supported, the paper would offer a useful empirical reference for QML-based intrusion detection in UAV swarms, spanning three QML paradigms, a public dataset, and explicit resource-footprint reporting. Its strengths include the concrete 8-feature engineering, the comparison of multiple quantum circuit depths, and the release of data partitions. However, the reported evidence supports only a narrower claim: among the models listed in Table I, the Hybrid QNN-8L scores highest on four metrics. The broader claims about outperforming 'all classical contenders,' about CNN dominance, and about low-data advantages are not established by the experiments as described.

major comments (4)

[Abstract; Section IV; Table I] The abstract and conclusion state that 'CNNs dominate when abundant data offset their larger parameter count' and that the Hybrid QNN 'outperforms all other quantum and classical contenders.' However, Table I lists no CNN and no classical neural network of any kind; the only classical baseline is an SVM whose kernel and regularization are unspecified. The central comparative claim is therefore tested against a single classical model, and the CNN sentence has no supporting experiment. Add CNN/MLP baselines under the same feature/balancing protocol, or revise the claims to 'among the evaluated models.'
[Table I] All metrics in Table I are single-run point estimates with no error bars, confidence intervals, or repeated-seed experiments. The headline differences—e.g., HybridQNN-8L accuracy 0.948 versus HybridQNN-4L 0.933 and SVM 0.924—could be within noise, especially given finite-shot quantum sampling. Since 'best performance' is the paper's central result, report mean ± standard deviation over multiple random seeds and, ideally, a significance test or overlap of confidence intervals.
[Section III.A; Section III.B] The experimental protocol is underdetermined. The paper does not specify train/validation/test split sizes, the balanced subset ratio, the number of shots M for quantum circuits, optimizer and learning rate, number of epochs, regularization, or the noise model, despite the abstract's claim of 'identical preprocessing, balancing, and noise-model assumptions.' These details are necessary to verify that the comparison is actually 'identical' and to assess reproducibility. Provide the full protocol or point precisely to where it is defined in the released code.
[Abstract; Section III.B] The abstract claims that 'quantum kernels and QT-NNs excel in low-data, nonlinear regimes,' but no low-data experiment is reported. All results come from a single dataset size; there is no learning-curve analysis, training-set subsampling, or other experiment that varies data availability. The terms 'low-data' and 'nonlinear regimes' are not operationalized. Either add such experiments or remove this claim from the abstract.

minor comments (6)

[Title; Abstract] The spacing in 'UA V' (e.g., in the title and abstract) is a formatting artifact; use 'UAV.'
[Section III.A] The text says the dataset contains 21 features grouped into three categories, then says 'the original 22 raw flow attributes' were distilled. Please reconcile the feature count.
[Data Availability] The paper states that 'the complete codebase and dataset partitions are publicly released,' but the only URL given is for the cuTN-QSVM framework. Provide a direct link to the benchmark code and the exact dataset partitions used in this study.
[Table I] Rows QT-NN (4,4) and QT-NN (16,4) report specificity 0.000 and sensitivity 1.000, indicating degenerate all-positive classifiers. These rows should be discussed or excluded, and the cause (e.g., training collapse) should be explained.
[Section III.B] The SVM baseline is not described: no kernel type, regularization parameter, or scaling details are given. Add a sentence specifying the classical baseline configuration.
[Eq. (13)] The notation 'w = T(g(θ)) ∪ w_base' is nonstandard; define the union/concatenation operation explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an empirical benchmark; self-cited tooling and QT-NN theory are not load-bearing to the reported comparisons.

full rationale

The paper does not present a derivation chain whose conclusions are equivalent to its inputs. It is an empirical benchmark on the external UAVIDS-2025 dataset [38]. All models consume the same 8-feature representation and are compared in Table I by trained/tested accuracy, F1, sensitivity, and specificity. The central claim that the eight-layer Hybrid QNN performs best is a measured result, not a consequence of any definition or fitting procedure. No parameter is fitted to a subset and then renamed as a prediction; the QT-NN rows are experimental results, and the QT-NN compression bound in Section II.D is a motivating theoretical assertion, not used to compute Table I. Self-citations to cuTN-QSVM [27] and QT-NN frameworks [34]-[37] describe implementation choices and prior methods, but the benchmark itself is external to those papers: the comparison would stand or fall on the reported experiment. The absence of a CNN baseline and the unsupported abstract statement about CNNs dominating are correctness/completeness concerns, not circularity. Therefore no circular step can be exhibited, and the circularity score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim relies on the simulated dataset being representative, the hand-crafted feature reduction being sufficient, and the tensor-network simulator being accurate. No new physical entities are introduced. The free parameters are mostly design choices in feature engineering and balancing, not fitted to optimize the final metrics beyond standard hyperparameter selection.

free parameters (3)

PMR sliding window size (w) = 100 ms (stated)
Chosen by hand for the peak-to-mean burst ratio feature; affects the 8-feature representation.
Feature selection (8 features from 22) = 8 features
A modeling choice that may discard information; not justified with ablation.
Balanced subset ratio = not specified
The paper states a balanced subset is constructed but does not report the exact class proportions or how imbalance is handled in training vs test.

assumptions (4)

domain assumption The UAVIDS-2025 simulated corpus is representative of real UAV swarm network traffic.
Section III.A: all experiments use this dataset; conclusions assume transferability to real-world conditions.
domain assumption The 8-feature distilled representation retains sufficient discriminative information for intrusion detection.
Section III.A: the authors select 8 features without a formal feature importance analysis or comparison against the full 22-feature set.
domain assumption Tensor-network simulation of quantum circuits faithfully approximates the behavior of ideal or noisy quantum devices.
Data Availability: results obtained via the cuTN-QSVM framework; no hardware or exact statevector verification is provided.
standard math The parameter-shift rule provides unbiased gradient estimates for the QNN and QT-NN training.
Section II.C: used in Eq. (12) and (14), standard result in variational quantum algorithms.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Quantum Machine Learning for UAV Swarm Intrusion Detection." pith.science (2026). https://pith.science/paper/JYUM5PFV

@misc{pith2026250901812,
  author       = {Pith},
  title        = {Pith review of: Quantum Machine Learning for UAV Swarm Intrusion Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JYUM5PFV}},
  note         = {Machine review of arXiv:2509.01812}
}

read the original abstract

Intrusion detection in unmanned-aerial-vehicle (UAV) swarms is complicated by high mobility, non-stationary traffic, and severe class imbalance. Leveraging a 120 k-flow simulation corpus that covers five attack types, we benchmark three quantum-machine-learning (QML) approaches - quantum kernels, variational quantum neural networks (QNNs), and hybrid quantum-trained neural networks (QT-NNs) - against strong classical baselines. All models consume an 8-feature flow representation and are evaluated under identical preprocessing, balancing, and noise-model assumptions. We analyse the influence of encoding strategy, circuit depth, qubit count, and shot noise, reporting accuracy, macro-F1, ROC-AUC, Matthews correlation, and quantum-resource footprints. Results reveal clear trade-offs: quantum kernels and QT-NNs excel in low-data, nonlinear regimes, while deeper QNNs suffer from trainability issues, and CNNs dominate when abundant data offset their larger parameter count. The complete codebase and dataset partitions are publicly released to enable reproducible QML research in network security.

Figures

Figures reproduced from arXiv: 2509.01812 by the authors.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

38 extracted references · 33 canonical work pages

[1]

Uav swarm commu- nication and control architectures: a review,

M. Campion, P. Ranganathan, and S. Faruque, “Uav swarm commu- nication and control architectures: a review,” Journal of Unmanned Vehicle Systems, vol. 7, no. 2, pp. 93–106, 2018

2018
[2]

Wireless communications and control for swarms of cellular- connected uavs,

T. Zeng, M. Mozaffari, O. Semiari, W. Saad, M. Bennis, and M. Deb- bah, “Wireless communications and control for swarms of cellular- connected uavs,” in 2018 52nd Asilomar Conference on Signals, Systems, and Computers , pp. 719–723, IEEE, 2018

work page 2018
[3]

Review of unmanned aerial vehicle swarm communication architectures and routing protocols,

X. Chen, J. Tang, and S. Lao, “Review of unmanned aerial vehicle swarm communication architectures and routing protocols,” Applied Sciences, vol. 10, no. 10, p. 3661, 2020

work page 2020
[4]

Swarmcontrol: An automated dis- tributed control framework for self-optimizing drone networks,

L. Bertizzolo, S. D’oro, L. Ferranti, L. Bonati, E. Demirors, Z. Guan, T. Melodia, and S. Pudlewski, “Swarmcontrol: An automated dis- tributed control framework for self-optimizing drone networks,” in IEEE INFOCOM 2020-IEEE Conference on Computer Communica- tions, pp. 1768–1777, IEEE, 2020

work page 2020
[5]

Survey on uav cellular communications: Prac- tical aspects, standardization advancements, regulation, and security challenges,

A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. Garcia- Rodriguez, and J. Yuan, “Survey on uav cellular communications: Prac- tical aspects, standardization advancements, regulation, and security challenges,” IEEE Communications surveys & tutorials , vol. 21, no. 4, pp. 3417–3442, 2019

work page 2019
[6]

Uav-enabled intelligent transportation systems for the smart city: Applications and challenges,

H. Menouar, I. Guvenc, K. Akkaya, A. S. Uluagac, A. Kadri, and A. Tuncer, “Uav-enabled intelligent transportation systems for the smart city: Applications and challenges,” IEEE Communications Mag- azine, vol. 55, no. 3, pp. 22–28, 2017

work page 2017
[7]

Flying ad-hoc networks (fanets): A survey,

I. Bekmezci, O. K. Sahingoz, and S ¸. Temel, “Flying ad-hoc networks (fanets): A survey,” Ad Hoc Networks , vol. 11, no. 3, pp. 1254–1270, 2013

2013
[8]

Micro aerial vehicle networks: An exper- imental analysis of challenges and opportunities,

M. Asadpour, B. Van den Bergh, D. Giustiniano, K. A. Hummel, S. Pollin, and B. Plattner, “Micro aerial vehicle networks: An exper- imental analysis of challenges and opportunities,” IEEE Communica- tions Magazine, vol. 52, no. 7, pp. 141–149, 2014

work page 2014

Show all 38 references

[9]

Low-altitude unmanned aerial vehicles-based internet of things services: Comprehensive survey and future perspectives,

N. H. Motlagh, T. Taleb, and O. Arouk, “Low-altitude unmanned aerial vehicles-based internet of things services: Comprehensive survey and future perspectives,” IEEE internet of things journal , vol. 3, no. 6, pp. 899–922, 2016

2016
[10]

Intrusion detection systems for networked unmanned aerial vehicles: A survey,

G. Choudhary, V . Sharma, I. You, K. Yim, R. Chen, and J.-H. Cho, “Intrusion detection systems for networked unmanned aerial vehicles: A survey,” in 2018 14th International Wireless Communications & Mobile Computing Conference (IWCMC) , pp. 560–565, IEEE, 2018

2018
[11]

A survey on security of uav swarm networks: attacks and countermea- sures,

X. Wang, Z. Zhao, L. Yi, Z. Ning, L. Guo, F. R. Yu, and S. Guo, “A survey on security of uav swarm networks: attacks and countermea- sures,” ACM Computing Surveys , vol. 57, no. 3, pp. 1–37, 2024

2024
[12]

Towards secure wireless mesh networks for uav swarm connectivity: Current threats, research, and opportunities,

M. A. Lopez, M. Baddeley, W. T. Lunardi, A. Pandey, and J.-P. Giacalone, “Towards secure wireless mesh networks for uav swarm connectivity: Current threats, research, and opportunities,” in2021 17th International Conference on Distributed Computing in Sensor Systems (DCOSS), p...

2021
[13]

A flow-based intrusion detection framework for internet of things networks,

L. Santos, R. Gonc ¸alves, C. Rabadao, and J. Martins, “A flow-based intrusion detection framework for internet of things networks,” Cluster Computing, vol. 26, no. 1, pp. 37–57, 2023

2023
[14]

Federated learning for uav swarms under class imbalance and power consumption constraints,

I. Mrad, L. Samara, A. A. Abdellatif, A. Al-Abbasi, R. Hamila, and A. Erbad, “Federated learning for uav swarms under class imbalance and power consumption constraints,” in 2021 IEEE Global Communi- cations Conference (GLOBECOM) , pp. 01–06, IEEE, 2021

2021
[15]

Challenges and opportunities in quantum optimization,

A. Abbas, A. Ambainis, B. Augustino, A. B ¨artschi, H. Buhrman, C. Coffrin, G. Cortiana, V . Dunjko, D. J. Egger, B. G. Elmegreen, et al., “Challenges and opportunities in quantum optimization,” Nature Reviews Physics, pp. 1–18, 2024

2024
[16]

Practical quantum advantage in quantum simulation,

A. J. Daley, I. Bloch, C. Kokail, S. Flannigan, N. Pearson, M. Troyer, and P. Zoller, “Practical quantum advantage in quantum simulation,” Nature, vol. 607, no. 7920, pp. 667–676, 2022

2022
[17]

Challenges and opportunities in quantum machine learning,

M. Cerezo, G. Verdon, H.-Y . Huang, L. Cincio, and P. J. Coles, “Challenges and opportunities in quantum machine learning,” Nature computational science, vol. 2, no. 9, pp. 567–576, 2022

2022
[18]

Exploring chemical space with chemistry-inspired dynamic quantum circuits in the nisq era,

L.-Y . Chen, T.-Y . Li, Y .-P. Li, N.-Y . Chen, and F. You, “Exploring chemical space with chemistry-inspired dynamic quantum circuits in the nisq era,” Journal of Chemical Theory and Computation , 2025

2025
[19]

Resource-efficient compilation of distributed quantum circuits for solving large-scale wireless communication network problems,

K.-C. Chen, F. Burt, S. Yu, C.-Y . Liu, M.-H. Hsieh, and K. K. Leung, “Resource-efficient compilation of distributed quantum circuits for solving large-scale wireless communication network problems,” in 2025 IEEE International Symposium on Circuits and Systems (ISCAS) , pp. 1–...

2025
[20]

Quantum-enhanced support vector machine for large-scale multi- class stellar classification,

K.-C. Chen, X. Xu, H. Makhanov, H.-H. Chung, and C.-Y . Liu, “Quantum-enhanced support vector machine for large-scale multi- class stellar classification,” in International Conference on Intelligent Computing, pp. 155–168, Springer, 2024

2024
[21]

Variational quantum circuits for deep reinforcement learning,

S. Y .-C. Chen, C.-H. H. Yang, J. Qi, P.-Y . Chen, X. Ma, and H.-S. Goan, “Variational quantum circuits for deep reinforcement learning,” IEEE access, vol. 8, pp. 141007–141024, 2020

2020
[22]

Quantum long short-term memory,

S. Y .-C. Chen, S. Yoo, and Y .-L. L. Fang, “Quantum long short-term memory,” in ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. 8622–8626, IEEE, 2022

2022
[23]

Federated quantum machine learning,

S. Y .-C. Chen and S. Yoo, “Federated quantum machine learning,” Entropy, vol. 23, no. 4, p. 460, 2021

2021
[24]

Compressedmediq: Hybrid quantum machine learning pipeline for high-dimensional neuroimaging data,

K.-C. Chen, Y .-T. Li, T.-Y . Li, C.-Y . Liu, P.-H. H. Lee, and C.-Y . Chen, “Compressedmediq: Hybrid quantum machine learning pipeline for high-dimensional neuroimaging data,” in 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW...

2025
[25]

Consensus-based distributed quantum kernel learning for speech recognition,

K.-C. Chen, W. Ma, and X. Xu, “Consensus-based distributed quantum kernel learning for speech recognition,” in 2025 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW), pp. 1–5, IEEE, 2025

2025
[26]

Quantum- classical-quantum workflow in quantum-hpc middleware with gpu acceleration,

K.-C. Chen, X. Li, X. Xu, Y .-Y . Wang, and C.-Y . Liu, “Quantum- classical-quantum workflow in quantum-hpc middleware with gpu acceleration,” in 2024 International Conference on Quantum Commu- nications, Networking, and Computing (QCNC) , pp. 304–311, IEEE, 2024

2024
[27]

Validating large-scale quantum machine learning: Efficient simulation of quantum support vector machines using tensor networks,

K.-C. Chen, T.-Y . Li, Y .-Y . Wang, S. See, C.-C. Wang, R. Wille, N.- Y . Chen, A.-C. Yang, and C.-Y . Lin, “Validating large-scale quantum machine learning: Efficient simulation of quantum support vector machines using tensor networks,” Machine Learning: Science and Technolo...

2025
[28]

Towards provably efficient quantum algorithms for large- scale machine-learning models,

J. Liu, M. Liu, J.-P. Liu, Z. Ye, Y . Wang, Y . Alexeev, J. Eisert, and L. Jiang, “Towards provably efficient quantum algorithms for large- scale machine-learning models,” Nature Communications , vol. 15, no. 1, p. 434, 2024

2024
[29]

Robust decentralized quantum kernel learning for noisy and adversarial environment,

W. Ma, K.-C. Chen, S. Yu, M. Liu, and R. Deng, “Robust decentralized quantum kernel learning for noisy and adversarial environment,” arXiv preprint arXiv:2504.13782, 2025

2025 arXiv
[30]

Noise- aware detectable byzantine agreement for consensus-based distributed quantum computing,

K.-C. Chen, M. Prest, F. Burt, S. Yu, and K. K. Leung, “Noise- aware detectable byzantine agreement for consensus-based distributed quantum computing,” in 2025 International Conference on Quantum Communications, Networking, and Computing (QCNC) , pp. 210–215, IEEE, 2025

2025
[31]

Quantum kernel-based long short-term memory,

Y .-C. Hsu, T.-Y . Li, and K.-C. Chen, “Quantum kernel-based long short-term memory,” in2025 IEEE International Conference on Acous- tics, Speech, and Signal Processing Workshops (ICASSPW) , pp. 1–5, IEEE, 2025

2025
[32]

The power of quantum neural networks,

A. Abbas, D. Sutter, C. Zoufal, A. Lucchi, A. Figalli, and S. Woerner, “The power of quantum neural networks,” Nature Computational Science, vol. 1, no. 6, pp. 403–409, 2021

2021
[33]

Shedding light on the future: Exploring quantum neural networks through optics,

S. Yu, Z. Jia, A. Zhang, E. Mer, Z. Li, V . Crescimanna, K.-C. Chen, R. B. Patel, I. A. Walmsley, and D. Kaszlikowski, “Shedding light on the future: Exploring quantum neural networks through optics,” Advanced Quantum Technologies, p. 2400074, 2024

2024
[34]

Quantum-train: Rethinking hybrid quantum- classical machine learning in the model compression perspective,

C.-Y . Liu, E.-J. Kuo, C.-H. A. Lin, J. G. Young, Y .-J. Chang, M.-H. Hsieh, and H.-S. Goan, “Quantum-train: Rethinking hybrid quantum- classical machine learning in the model compression perspective,” Quantum Machine Intelligence , vol. 7, no. 2, p. 80, 2025

2025
[35]

Quantum-train with tensor network mapping model and distributed circuit ansatz,

C.-Y . Liu, C.-H. A. Lin, and K.-C. Chen, “Quantum-train with tensor network mapping model and distributed circuit ansatz,” in ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pp. 1–4, IEEE, 2025

2025
[36]

Programming variational quantum circuits with quantum-train agent,

C.-Y . Liu, S. Y .-C. Chen, K.-C. Chen, W.-J. Huang, and Y .-J. Chang, “Programming variational quantum circuits with quantum-train agent,” in 2025 International Conference on Quantum Communications, Net- working, and Computing (QCNC) , pp. 544–548, IEEE, 2025

2025
[37]

Quantum- train-based distributed multi-agent reinforcement learning,

K.-C. Chen, S. Y .-C. Chen, C.-Y . Liu, and K. K. Leung, “Quantum- train-based distributed multi-agent reinforcement learning,” in 2025 IEEE Symposium for Multidisciplinary Computational Intelligence Incubators (MCII Companion) , pp. 1–5, IEEE, 2025

2025
[38]

Uavids-2025: A bench- mark dataset for intrusion detection in uav networks using machine learning techniques,

Q. Zeng, A. Bashir, and F. Nait-Abdesselam, “Uavids-2025: A bench- mark dataset for intrusion detection in uav networks using machine learning techniques,” in 2025 IEEE Conference on Communications and Network Security (CNS) , IEEE, 2025

2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Uav swarm commu- nication and control architectures: a review,

M. Campion, P. Ranganathan, and S. Faruque, “Uav swarm commu- nication and control architectures: a review,” Journal of Unmanned Vehicle Systems, vol. 7, no. 2, pp. 93–106, 2018

2018

[2] [2]

Wireless communications and control for swarms of cellular- connected uavs,

T. Zeng, M. Mozaffari, O. Semiari, W. Saad, M. Bennis, and M. Deb- bah, “Wireless communications and control for swarms of cellular- connected uavs,” in 2018 52nd Asilomar Conference on Signals, Systems, and Computers , pp. 719–723, IEEE, 2018

work page 2018

[3] [3]

Review of unmanned aerial vehicle swarm communication architectures and routing protocols,

X. Chen, J. Tang, and S. Lao, “Review of unmanned aerial vehicle swarm communication architectures and routing protocols,” Applied Sciences, vol. 10, no. 10, p. 3661, 2020

work page 2020

[4] [4]

Swarmcontrol: An automated dis- tributed control framework for self-optimizing drone networks,

L. Bertizzolo, S. D’oro, L. Ferranti, L. Bonati, E. Demirors, Z. Guan, T. Melodia, and S. Pudlewski, “Swarmcontrol: An automated dis- tributed control framework for self-optimizing drone networks,” in IEEE INFOCOM 2020-IEEE Conference on Computer Communica- tions, pp. 1768–1777, IEEE, 2020

work page 2020

[5] [5]

Survey on uav cellular communications: Prac- tical aspects, standardization advancements, regulation, and security challenges,

A. Fotouhi, H. Qiang, M. Ding, M. Hassan, L. G. Giordano, A. Garcia- Rodriguez, and J. Yuan, “Survey on uav cellular communications: Prac- tical aspects, standardization advancements, regulation, and security challenges,” IEEE Communications surveys & tutorials , vol. 21, no. 4, pp. 3417–3442, 2019

work page 2019

[6] [6]

Uav-enabled intelligent transportation systems for the smart city: Applications and challenges,

H. Menouar, I. Guvenc, K. Akkaya, A. S. Uluagac, A. Kadri, and A. Tuncer, “Uav-enabled intelligent transportation systems for the smart city: Applications and challenges,” IEEE Communications Mag- azine, vol. 55, no. 3, pp. 22–28, 2017

work page 2017

[7] [7]

Flying ad-hoc networks (fanets): A survey,

I. Bekmezci, O. K. Sahingoz, and S ¸. Temel, “Flying ad-hoc networks (fanets): A survey,” Ad Hoc Networks , vol. 11, no. 3, pp. 1254–1270, 2013

2013

[8] [8]

Micro aerial vehicle networks: An exper- imental analysis of challenges and opportunities,

M. Asadpour, B. Van den Bergh, D. Giustiniano, K. A. Hummel, S. Pollin, and B. Plattner, “Micro aerial vehicle networks: An exper- imental analysis of challenges and opportunities,” IEEE Communica- tions Magazine, vol. 52, no. 7, pp. 141–149, 2014

work page 2014