Pith. sign in

REVIEW 4 major objections 7 minor 38 references

A hybrid quantum and quantum-inspired framework estimates and optimizes Quality of Experience in cellular networks, matching classical accuracy while cutting model load times and optimizer time-to-solution.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Quantum-inspired regressors match classical ML for cloud gaming KQI prediction on a controlled testbed, and a tensor-network optimizer matches brute-force with a modest speedup.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Credible proof-of-concept for quantum-inspired ML in O-RAN, but the QML claim is unverifiable without disclosing the QUBO solver. the 4 major comments →

arxiv 2509.01008 v1 pith:D3TJ7SOK submitted 2025-08-31 cs.NI cs.LGquant-ph

Quantum-based QoE Optimization in Advanced Cellular Networks: Integration and Cloud Gaming Use Case

Fatma Chaouech , Javier Villegas , Ant\'onio Pereira , Carlos Baena , Sergio Fortes , Raquel Barco , Dominic Gribben , Mohammad Dib
show 3 more authors
Alba Villarino Aser Cortines Rom\'an Or\'us
This is my paper
classification cs.NI cs.LGquant-ph
keywords quantum machine learningquantum-inspired algorithmsquality of experiencecloud gamingO-RANQUBOtensor networksnetwork optimization
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 sets out to show that quantum-based and quantum-inspired machine learning can be integrated into real cellular network management without waiting for fault-tolerant quantum hardware. It constructs a two-phase pipeline—offline training, online deployment inside the O-RAN architecture—in which an estimator converts user and network metrics into a predicted Quality of Experience (QoE) score, and an optimizer picks the cell/service configuration that best balances that score against resource use. In the cloud gaming case study, the SML estimator (a QBoost ensemble) delivers MASE values of 0.539 (latency), 0.135 (freeze percentage), and 0.076 (effective FPS), matching or beating the best reproducible classical baselines (0.538, 0.139, 0.084) while loading in 0.5–0.7 ms against 7.9–8.0 ms for classical models. The STN/TTOpt optimizer finds the same optimum as brute-force search in most tested configurations and shortens average time-to-solution by roughly 15%. The authors conclude that such hybrid methods are a practical near-term route to quantum-influenced network intelligence, while acknowledging that the small, controlled dataset limits the strength of the comparison.

Core claim

On the paper's own terms, the central discovery is that QML and QI estimators can be both smaller and faster than classical baselines while keeping comparable or better accuracy. The SML regressor—a QBoost ensemble whose weak-learner weights come from solving a QUBO—achieves the best MASE for effective FPS (0.076 versus 0.084 for the classical KNR) and loads in 0.5 ms, roughly an order of magnitude faster than the classical models; the STN tensor-network regressor posts the lowest freeze-percentage error (0.119 MASE) at 13 features. The TTOpt optimizer, using the same tensor-train machinery, recovers the brute-force optimum for nearly every tested resolution and α trade-off and, in the repre

What carries the argument

The argument is carried by two mechanisms. First, QBoost, an ensemble method that converts the selection and weighting of weak learners into a Quadratic Unconstrained Binary Optimization (QUBO) problem—a binary optimization with up to quadratic terms that quantum annealers are designed to solve. The trained model is a compact set of weighted learners, which explains the sub-millisecond loading times. Second, the tensor-train (TT) format shared by the STN regressor and the TTOpt optimizer: a low-rank tensor network that approximates high-dimensional functions using a chain of smaller tensors, with the maximum-volume (MaxVol) principle choosing which grid points to evaluate. The TT machinery i

Load-bearing premise

The load-bearing premise is that the SML estimator's training truly used a quantum annealer; the paper never says what solved the optimization problem, so if an ordinary classical algorithm produced the model weights, the quantum explanation for the results is unsupported.

What would settle it

Look at the SML training configuration or logs: if the QUBO was minimized by a classical solver (simulated annealing, tabu search, or a standard optimizer) rather than a quantum annealer, the paper's quantum claim for SML is void. Alternatively, retrain SML on the same cloud-gaming data with a real quantum annealer and with a purely classical QUBO solver; if both produce models with the same accuracy and loading time, then 'quantum advantage' in training is not empirically established.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Deploying the QoE estimator and optimizer as O-RAN rApps/xApps becomes realistic: sub-millisecond model loading fits the Near-RT RIC's 10 ms–1 s control-loop budget.
  • Operators can automate QoE-driven configuration for cloud gaming sessions—choosing PRB allocation, resolution, and frame rate—without exhaustive search, since TTOpt recovers near-optimal solutions faster than brute force.
  • SML's error keeps falling as the feature count grows, so the method is a candidate for richer network datasets where classical baselines saturate.
  • Because trained SML models run inference on classical hardware, the quantum requirement is confined to training, which lowers the barrier to near-term deployment.

Where Pith is reading between the lines

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

  • Editorial inference: the paper never identifies the solver that minimized the QUBO. If it was a classical optimizer (simulated annealing, tabu search, etc.) rather than a quantum annealer, the reported small-model advantage is a property of the QBoost formulation, not of quantum hardware, and the 'advantage of using QC for training' quoted in Section V-B is unsubstantiated.
  • Editorial inference: a head-to-head against classical sparse boosting (e.g., L1-regularized AdaBoost) on the same data would reveal whether SML's sub-millisecond loading and accuracy come from quantum annealing or from enforcing sparse weights. This test is not in the paper.
  • Editorial inference: the 15% time-to-solution reduction is measured on a 3-dimensional search space (PRB × resolution × FPS) that brute force can fully enumerate. The meaningful extension—which the authors point toward but do not run—is to test TTOpt on configuration spaces with too many dimensions for enumeration, comparing against classical optimizers such as Bayesian optimization.
  • Editorial inference: the authors' own limitation note that classical methods 'also obtain great results' on this dataset. The projected advantage for higher-dimensional data is therefore a hypothesis, not a demonstrated result; a falsifiable version would need a dataset where the classical baselines' MASE plateaus at high feature counts while SML keeps improving.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 7 minor

Summary. The paper proposes a hybrid quantum/quantum-inspired framework for QoE estimation and optimization in cellular networks, integrated into the O-RAN architecture as Q-rApps/Q-xApps. The estimator uses two proprietary methods—SML (claimed to be a QBoost-based quantum-annealed ensemble regressor) and STN (a tensor-network regressor)—to predict cloud-gaming KQIs (latency, freeze percentage, EFPS) from a public 4G/5G dataset, and compares them with classical SVR, RF, and KNR baselines. The optimizer uses STN/TTOpt to select PRB allocation, resolution, and FPS by maximizing a hand-crafted objective J, and is compared with brute-force search. The paper reports that QML/QI models achieve similar or better MASE than classical models, that SML has sub-millisecond loading times, and that the optimizer matches brute-force optima in most cases while reducing time-to-solution by about 15%. The authors acknowledge that the dataset is limited and that classical methods also perform well.

Significance. If the quantum attribution of SML is substantiated, the paper is a useful integration and benchmarking study for QML/QI in network management, with a concrete O-RAN deployment path, a publicly available dataset, and clearly defined KQIs and metrics. The use of 5-fold cross-validation, 100 optimizer runs, and comparison with reproducible classical baselines are strengths. However, the central contribution depends on an undisclosed training solver for the claimed quantum-enhanced method, and the reported accuracy advantages are small and not statistically characterized. The paper is therefore significant mainly as a proof-of-concept, and its headline claims require substantially stronger support before they can be accepted.

major comments (4)
  1. [§IV.B.1 and §V.B] The identification of SML as a quantum method is not verifiable. The text states that SQER is derived from QBoost and that training is performed via quantum annealing, but the manuscript never identifies the solver: no QPU (e.g., D-Wave), no simulator, no classical QUBO fallback, and no annealing parameters. If the QUBO was optimized by a classical method, SML is a classical ensemble and the claim in §V.B of 'the advantage of using QC for training' is unsupported. This is load-bearing for the abstract's QML claim, so the solver and execution mode must be disclosed, with evidence that annealing was actually used.
  2. [§V.A, Figure 3, Table 1] The feature-count selection is circular. The best number of features for each model is read from the same MASE-versus-features curves that are then reported as the headline results, with no held-out model-selection step or nested cross-validation. The MASE differences are tiny (e.g., Latency: STN 0.529, SVR 0.538, SML 0.539; Freeze: SML 0.135 vs RF 0.139; EFPS: SML 0.076 vs KNR 0.084), and no confidence intervals or significance tests are given. There is also an internal inconsistency: §V.A.1 states SML's best Latency MASE is 0.535 at 9 features, while Table 1 reports 0.539 for the same configuration; with SVR at 0.538, the claim of quantum-based superiority for Latency is not supported. This undermines the central estimation-accuracy claim.
  3. [§V.B and Table 1] The timing comparison is not apples-to-apples. Table 1 reports each model's best feature count (e.g., SML 9/9/13, STN 8/13/10, SVR 10, RF 9, KNR 5), but §V.B states that 'for simplicity, only the models with 13 features were evaluated for each KQI.' Thus the SML loading times in Table 1 are for 13-feature models, while the classical baselines' best configurations use fewer features. The sub-millisecond loading advantage may therefore reflect model size or feature count rather than any property of quantum training. The timing comparison should use matched feature counts and model sizes, and report run-to-run variability.
  4. [§V.C and Table 2] The optimization results lack statistical characterization and scaling evidence. The algorithm was run 100 times, but no variance, standard deviation, or failure rate is reported; the time-to-solution differences (e.g., 12.6 ms vs 10.5 ms) are small and hardware-dependent. Since brute-force exhaustive search over the small grid is feasible, the claim that STN is 'a promising candidate for scaling to higher-dimensional problems' is not supported by the presented experiments. Reporting the number of function evaluations and a scaling study with larger search spaces would be needed to justify that conclusion.
minor comments (7)
  1. [§IV.A.2] The outlier detection and removal procedure is not specified (method, threshold, or number of removed samples). The reduction from 3,694 cloud-gaming entries to 3,467 samples should be explained.
  2. [§IV.A.3] It is unclear whether the mutual-information feature selection was computed on the full dataset or only on the training split. If the former, there is information leakage from the test set into feature selection.
  3. [Eq. (5)] Equation (5) contains formatting errors: '[efps' and '\latency' are malformed, and the expression appears to be missing a closing brace. Please correct the notation.
  4. [Table 2] In the 1080p, α=0.8 row, the STN Optimizer column appears garbled and duplicated ('26 1440p 120 75.45 118 10.51529 4k 120 95.75 125'). This must be corrected.
  5. [§V.B] The timing evaluation is described as averaging over 100 runs on an Apple M1 Pro Docker instance, but no standard deviations or environment details are provided. This makes the 15% optimizer speedup difficult to interpret.
  6. [References] Reference [16] is missing its title, and reference [29] has an unusual quoted title. Please check the reference formatting.
  7. [Conclusions] The conclusion that 'the quantum-based approaches obtained lower estimation errors' should be qualified: for Latency, SML's MASE (0.539) is slightly worse than SVR's (0.538), and only STN is lower.

Circularity Check

0 steps flagged

No significant circularity: estimator accuracy is benchmarked against a naive predictor on external public data, and the optimizer against exhaustive search; only minor self-citations and a disclosure gap remain.

full rationale

The paper's central numerical claims are self-contained rather than circular. The estimator (SML/STN) is evaluated via MASE, which is defined relative to a naive estimator (Sec. IV.A.4), on the external public dataset of [21]/[22]; the classical baselines are taken from the prior study [35] but are retrained/reproduced in this work, so the comparison is not forced by construction. The optimizer (TTOpt) is benchmarked against a brute-force search on the paper's explicitly stated objective J (Eqs. 4-6); matching the brute-force optimum is an empirical property of the algorithm, not a restatement of the input. Self-citations to [18], [33], [35] are architectural or baseline references and are not load-bearing for the accuracy or timing results. Two caveats are flagged but are not circularity: (i) the paper does not disclose whether the QUBO underlying SQER was solved on a real annealer or by simulation, so the 'quantum' attribution in Sec. IV.B.1 and the claim of 'advantage of using QC for training' in Sec. V.B rest on an unverified premise; (ii) the feature counts in Fig. 3/Table 1 are selected from the evaluation curves, which can bias the reported minima, but this is a data-selection concern rather than a derivation that equates output to input. The limitations in Sec. VI also acknowledge the low complexity of the dataset, further reducing the strength of the claims without making them circular.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

The central comparison rests on a few hand-chosen constants (cost function parameters, feature counts, discretization resolution) and on assumptions imported from prior work. The independent evidence is the public dataset and the naive MASE baseline, but the feature counts and cost constants are not externally benchmarked. No new physical entities are introduced.

free parameters (4)
  • Hand-picked cost-function constants in Eqs. (5)-(6) = 43.25, 150, 53, 10, MinRes
    Chosen ad hoc to model diminishing returns for EFPS, latency, and PRB cost; these constants shape the optimizer objective and are not derived from data or external benchmarks.
  • Number of features per KQI model = 5 to 13 depending on KQI and model
    Selected from the MASE versus number-of-features curves in Figure 3 and reported in Table 1; this is model selection done on the same evaluation curves that are later cited as performance.
  • STN discretization resolution n_allowed_vals = 100 for continuous features
    Fixed without sensitivity analysis; affects the granularity of the tensor train input and may influence accuracy.
  • Outlier removal threshold = Not stated
    Preprocessing is described as 'outliers were detected and removed through statistical analysis' without specifying the criterion or the number of removed samples.
axioms (5)
  • standard math QBoost regression QUBO encoding of weak-learner weights is valid and solvable by annealing
    Adopted from refs [24,27,28]; the SML method relies entirely on this encoding.
  • standard math Tensor train decomposition can approximate the exponential interaction space of features
    Adopted from Exponential Machines [29] and TTOpt [30]; the STN regressor and optimizer are built on this assumption.
  • domain assumption The controlled-testbed dataset [21] is representative of cloud gaming QoE for evaluating generalization
    Section IV.A.1 states all samples were gathered under a standardized service configuration and execution framework in a 4G/5G testbed; the conclusions generalize only if this dataset is a valid proxy for real-world conditions.
  • domain assumption Mutual information computed on the available features reflects predictive relevance for KQIs
    Feature ranking in Section IV.A.3 assumes MI is a sufficient criterion for selecting the 13 features; no ablation against other feature selection methods is provided.
  • ad hoc to paper The hand-crafted optimizer objective J in Eq. (4) with exponential cost functions captures the operator's QoE-versus-resource trade-off
    Eqs. (5) and (6) are invented for this paper with chosen constants; the optimization results are only meaningful if this objective reflects real network priorities.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum-based QoE Optimization in Advanced Cellular Networks: Integration and Cloud Gaming Use Case." pith.science (2026). https://pith.science/paper/D3TJ7SOK

@misc{pith2026250901008,
  author       = {Pith},
  title        = {Pith review of: Quantum-based QoE Optimization in Advanced Cellular Networks: Integration and Cloud Gaming Use Case},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/D3TJ7SOK}},
  note         = {Machine review of arXiv:2509.01008}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

This work explores the integration of Quantum Machine Learning (QML) and Quantum-Inspired (QI) techniques for optimizing end-to-end (E2E) network services in telecommunication systems, particularly focusing on 5G networks and beyond. The application of QML and QI algorithms is investigated, comparing their performance with classical Machine Learning (ML) approaches. The present study employs a hybrid framework combining quantum and classical computing leveraging the strengths of QML and QI, without the penalty of quantum hardware availability. This is particularized for the optimization of the Quality of Experience (QoE) over cellular networks. The framework comprises an estimator for obtaining the expected QoE based on user metrics, service settings, and cell configuration, and an optimizer that uses the estimation to choose the best cell and service configuration. Although the approach is applicable to any QoE-based network management, its implementation is particularized for the optimization of network configurations for Cloud Gaming services. Then, it is evaluated via performance metrics such as accuracy and model loading and inference times for the estimator, and time to solution and solution score for the optimizer. The results indicate that QML models achieve similar or superior accuracy to classical ML models for estimation, while decreasing inference and loading times. Furthermore, potential for better performance is observed for higher-dimensional data, highlighting promising results for higher complexity problems. Thus, the results demonstrate the promising potential of QML in advancing network optimization, although challenges related to data availability and integration complexities between quantum and classical ML are identified as future research lines.

Figures

Figures reproduced from arXiv: 2509.01008 by Alba Villarino, Ant\'onio Pereira, Aser Cortines, Carlos Baena, Dominic Gribben, Fatma Chaouech, Javier Villegas, Mohammad Dib, Raquel Barco, Rom\'an Or\'us, Sergio Fortes.

Figure 1
Figure 1. Figure 1: FIGURE 1: Network architecture and integration of QML/QI ML algorithms for a QoE Estimation and Optimization. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIGURE 2: Mutual information scores for each CG per [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIGURE 3: MASE evolution of different models predicting [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIGURE 4: TTOpt vs Brute-force: Average objective values for different resolution requirements and alpha parameters. [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗

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

38 extracted references · 29 canonical work pages · 3 internal anchors

  1. [1]

    Bottleneck identification and failure prevention with procedural learning in 5G RAN,

    T. Sundqvist, M. Bhuyan, and E. Elmroth, “Bottleneck identification and failure prevention with procedural learning in 5G RAN,” in 2023 IEEE/ACM 23rd International Symposium on Cluster, Cloud and Internet Computing (CCGrid) , 2023, pp. 426–436

    work page 2023

  2. [2]

    Failure Pre- diction in Cloud Native 5G Core With eBPF-based Observability,

    J. Kawasaki, D. Koyama, T. Miyasaka, and T. Otani, “Failure Pre- diction in Cloud Native 5G Core With eBPF-based Observability,” in 2023 IEEE 97th Vehicular Technology Conference (VTC2023-Spring), 2023, pp. 1–6

    work page 2023

  3. [3]

    Secure Handover in 5G-V2X: Detecting and Localizing DoS Attack to Ensure Reliable Communication,

    M. R. Dey and M. Patra, “Secure Handover in 5G-V2X: Detecting and Localizing DoS Attack to Ensure Reliable Communication,” IEEE Transactions on Intelligent Transportation Systems , pp. 1–14, 2025

    work page 2025

  4. [4]

    Resource allocation in multi-user cellular networks: A transformer-based deep reinforce- ment learning approach,

    Z. Di, Z. Zhong, Q. Pengfei, Q. Hao, and S. Bin, “Resource allocation in multi-user cellular networks: A transformer-based deep reinforce- ment learning approach,” China Communications , vol. 21, no. 5, pp. 77–96, 2024

    work page 2024

  5. [5]

    A Resource Allocation Method for Multi-Cell Cellular Networks with System Capacity and QoE Tradeoff,

    L. Zhang, X. Zhou, J. Zou, and X. Liao, “A Resource Allocation Method for Multi-Cell Cellular Networks with System Capacity and QoE Tradeoff,” in 2024 4th International Conference on Electronic Information Engineering and Computer Science (EIECS) , 2024, pp. 1087–1092

    work page 2024

  6. [6]

    Tensor networks for complex quantum systems,

    R. Or ´us, “Tensor networks for complex quantum systems,” Nature Reviews Physics , vol. 1, no. 9, pp. 538–550, Sep 2019. [Online]. Available: https://doi.org/10.1038/s42254-019-0086-7

    doi:10.1038/s42254-019-0086-7 2019

  7. [7]

    M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information: 10th Anniversary Edition . Cambridge University Press, 2010

    2010

  8. [8]

    Quantum Computing in the NISQ era and beyond,

    J. Preskill, “Quantum Computing in the NISQ era and beyond,” Quantum, vol. 2, p. 79, Aug. 2018. [Online]. Available: https: //doi.org/10.22331/q-2018-08-06-79

    doi:10.22331/q-2018-08-06-79 2018

  9. [9]

    Adiabatic quantum computation,

    T. Albash and D. A. Lidar, “Adiabatic quantum computation,” Rev. Mod. Phys. , vol. 90, p. 015002, Jan 2018. [Online]. Available: https://link.aps.org/doi/10.1103/RevModPhys.90.015002

    doi:10.1103/revmodphys.90.015002 2018

  10. [10]

    Ising formulations of many NP problems,

    A. Lucas, “Ising formulations of many NP problems,” Frontiers in Physics, vol. 2, 2014. [Online]. Available: https://www.frontiersin. org/journals/physics/articles/10.3389/fphy.2014.00005

    arXiv 2014

  11. [11]

    Quantum machine learning,

    J. Biamonte, P. Wittek, N. Pancotti, P. Rebentrost, N. Wiebe, and S. Lloyd, “Quantum machine learning,” Nature, vol. 549, no. 7671, pp. 195–202, Sep 2017. [Online]. Available: https: //doi.org/10.1038/nature23474

    doi:10.1038/nature23474 2017

  12. [12]

    Circuit-centric quantum classifiers,

    M. Schuld, A. Bocharov, K. M. Svore, and N. Wiebe, “Circuit-centric quantum classifiers,” Phys. Rev. A , vol. 101, p. 032308, Mar

    work page

  13. [13]

    ”tensor networks in machine learning

    R. Sengupta, S. Adhikary, I. Oseledets, and J. Biamonte, “”tensor networks in machine learning”,” Eur. Math. Soc. Mag. , no. 126, pp. 4–12, Oct. 2022. [Online]. Available: https://euromathsoc.org/ magazine/articles/101

    work page 2022

  14. [14]

    Quantum communication networks: Design, reliability, and security,

    B. Hildebrand, A. Ghimire, F. Amsaad, A. Razaque, and S. P. Mohanty, “Quantum communication networks: Design, reliability, and security,” IEEE Potentials, pp. 2–8, 2023

    work page 2023

  15. [15]

    Quantum cryptography: Public key distribution and coin tossing,

    C. H. Bennett and G. Brassard, “Quantum cryptography: Public key distribution and coin tossing,” Theoretical Computer Science , vol. 560, pp. 7–11, 2014, theoretical Aspects of Quantum Cryptography – celebrating 30 years of BB84. [Online]. Available: https: //www.sciencedirect.com/science/article/pii/S0304397514004241

    work page 2014

  16. [16]

    S. Kasi, P. Warburton, J. Kaewell, and K. Jamieson, “,” IEEE Trans- actions on Quantum Engineering , vol. 4, pp. 1–17, 2023

    work page 2023

  17. [17]

    Quantum Graph Neural Network for Resource Management in Wireless Communication,

    L. T. Giang, N. X. Tung, and W.-J. Hwang, “Quantum Graph Neural Network for Resource Management in Wireless Communication,” in 12 VOLUME , 2025 International Conference on Artificial Intelligence in Information and Communication (ICAIIC) , 2025, pp. 0128–0130

    work page 2025

  18. [18]

    Quantum- based Intelligence for Cellular Network Management: Use Cases, Challenges, and O-RAN Integration,

    S. Fortes, J. Villegas, F. Chaouech, A. Pereira, A. Villarino, A. Cortines, C. Baena, R. Or ´us, and R. Barco, “Quantum- based Intelligence for Cellular Network Management: Use Cases, Challenges, and O-RAN Integration,” Jun. 2025. [Online]. Available: http://dx.doi.org/10.36227/techrxiv.174952955.52927388/v1

    work page arXiv 2025

  19. [19]

    Exploring the potential advantages of quantum computing in telecommunication networks,

    A. J. Awan, M. A. Ullah, C. J. Yang, R. P. Sircar, B. Grafulla- Gonz´alez, and C. Granbom’, “Exploring the potential advantages of quantum computing in telecommunication networks,” Ericsson Technology Review, 2025

    work page 2025

  20. [20]

    Modeling of Key Quality Indicators for End-to-End Network Management: Preparing for 5G,

    A. Herrera-Garcia, S. Fortes, E. Baena, J. Mendoza, C. Baena, and R. Barco, “Modeling of Key Quality Indicators for End-to-End Network Management: Preparing for 5G,” IEEE Vehicular Technology Magazine, vol. 14, no. 4, pp. 76–84, 2019

    work page 2019

  21. [21]

    E2E dataset of Video Streaming and Cloud Gaming services over 4G and 5G,

    C. Baena, O. S. Pe ˜naherrera-Pulla, L. Camacho, R. Barco, and S. Fortes, “E2E dataset of Video Streaming and Cloud Gaming services over 4G and 5G,” IEEE Dataport, 2022. [Online]. Available: https://dx.doi.org/10.21227/k0w8-qz67

    work page doi:10.21227/k0w8-qz67 2022

  22. [22]

    Video Streaming and Cloud Gaming Services Over 4G and 5G: A Complete Network and Service Metrics Dataset,

    C. Baena, O. S. Pe ˜naherrera-Pulla, L. Camacho, R. Barco, and S. Fortes, “Video Streaming and Cloud Gaming Services Over 4G and 5G: A Complete Network and Service Metrics Dataset,” IEEE Communications Magazine, vol. 61, no. 9, pp. 154–160, 2023

    work page 2023

  23. [23]

    Information Theory,

    A. Orlitsky, “Information Theory,” in Encyclopedia of Physical Science and Technology (Third Edition) , 3rd ed., R. A. Meyers, Ed. New York: Academic Press, 2003, pp. 751–769. [Online]. Available: https: //www.sciencedirect.com/science/article/pii/B0122274105003379

    work page 2003

  24. [24]

    QBoost: Large Scale Classifier Training with Adiabatic Quantum Optimization,

    H. Neven, V . S. Denchev, G. Rose, and W. G. Macready, “QBoost: Large Scale Classifier Training with Adiabatic Quantum Optimization,” in Proceedings of the Asian Conference on Machine Learning, ser. Proceedings of Machine Learning Research, S. C. H. Hoi and W. Buntine, Eds., vol. 25. Singapore Management University, Singapore: PMLR, 04–06 Nov 2012, pp. 33...

    work page 2012

  25. [25]

    Training a Large Scale Classifier with the Quantum Adiabatic Algorithm

    H. Neven, V . S. Denchev, G. Rose, and W. G. Macready, “Training a Large Scale Classifier with the Quantum Adiabatic Algorithm,” 2009. [Online]. Available: https://arxiv.org/abs/0912.0779

    work page internal anchor Pith review Pith/arXiv arXiv 2009

  26. [26]

    A tutorial on formulating and using qubo models,

    F. Glover, G. Kochenberger, and Y . Du, “A tutorial on formulating and using qubo models,” 2019. [Online]. Available: https://arxiv.org/ abs/1811.11538

    Pith/arXiv arXiv 2019

  27. [27]

    QBoost for regression problems: solving partial differential equations

    C. B. D. G ´oes, T. O. Maciel, G. G. Pollachini, R. Cuenca, J. P. L. C. Salazar, and E. I. Duzzioni, “QBoost for regression problems: solving partial differential equations,” 2021. [Online]. Available: https://arxiv.org/abs/2108.13346

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  28. [28]

    Floating-Point Calculations on a Quantum Annealer: Division and Matrix Inversion

    M. L. Rogers and R. L. S. Jr, “Floating-Point Calculations on a Quantum Annealer: Division and Matrix Inversion,” 2019. [Online]. Available: https://arxiv.org/abs/1901.06526

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  29. [29]

    Exponential Machines,

    A. Novikov, M. Trofimov, and I. Oseledets, “Exponential Machines,”

    work page

  30. [30]

    TTOpt: A Maximum V olume Quantized Tensor Train-based Optimization and its Application to Reinforcement Learning,

    K. Sozykin, A. Chertkov, R. Schutski, A.-H. Phan, A. S. CICHOCKI, and I. Oseledets, “TTOpt: A Maximum V olume Quantized Tensor Train-based Optimization and its Application to Reinforcement Learning,” in Advances in Neural Information Processing Systems , S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh, Eds., vol. 35. Curran Associates, I...

    work page 2022

  31. [31]

    The maximal-volume concept in approximation by low-rank matrices,

    S. Goreinov and E. Tyrtyshnikov, “The maximal-volume concept in approximation by low-rank matrices,” Contemporary Mathematics , vol. 208, 01 2001

    work page 2001

  32. [32]

    Amarisoft,

    “Amarisoft,” https://www.amarisoft.com/technology/, visited in Febru- ary 2024

    work page 2024

  33. [33]

    A Framework to boost the potential of network-in-a-box solutions,

    C. Baena, S. Fortes, O. Pe ˜naherrera, and R. Barco, “A Framework to boost the potential of network-in-a-box solutions,” in 2021 12th International Conference on Network of the Future (NoF) , 2021, pp. 1–3

    work page 2021

  34. [34]

    Multiverse computing - singularity integration guide,

    IBM Quantum, “Multiverse computing - singularity integration guide,” 2024, accessed: 2025-06-01. [Online]. Available: https: //docs.quantum.ibm.com/guides/multiverse-computing-singularity

    work page 2024

  35. [35]

    Measuring and estimating Key Quality Indicators in Cloud Gaming services,

    C. Baena, O. Pe ˜naherrera-Pulla, R. Barco, and S. Fortes, “Measuring and estimating Key Quality Indicators in Cloud Gaming services,” Computer Networks , vol. 231, p. 109808, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S1389128623002530 Fatma Chaouech (fatma.chaouech@multiverse computing.com) received her M.Sc. in Softwa...

    work page 2023

  36. [2017]

    Available: https://arxiv.org/abs/1605.03795

    [Online]. Available: https://arxiv.org/abs/1605.03795

    Pith/arXiv arXiv

  37. [2020]

    Available: https://link.aps.org/doi/10.1103/PhysRevA

    [Online]. Available: https://link.aps.org/doi/10.1103/PhysRevA. 101.032308

    doi:10.1103/physreva

  38. [2021]

    She previously worked as a Quantum Software Engineer and is currently a Manager at Multiverse Computing

    She has over three years of experience in the field of quantum computing. She previously worked as a Quantum Software Engineer and is currently a Manager at Multiverse Computing. Aser Cortines (aser.cortines@multiversecompu ting.com) holds an M.Sc. from ´Ecole Polytech- nique and a Ph.D. in Applied Mathematics from the University of Paris Diderot. He held...

    work page 2022

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.