Pith. sign in

REVIEW 3 major objections 4 minor 20 references

A Non-Monotonic Relationship: An Empirical Analysis of Hybrid Quantum Classifiers for Unseen Ransomware Detection

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

Pith's one-line read A hybrid quantum classifier for unseen ransomware detection is beaten decisively by classical logistic regression, and its performance does not scale monotonically with qubit count.

desk verdict Honest, clearly reported negative-result benchmark for PCA+VQC on ransomware, but the paper's main analytical claim—non-monotonic trainability/barren plateaus—is confounded by unequal training budgets and a single seed. read the letter →

arxiv 2509.07924 v1 pith:7ROBMYUE submitted 2025-09-09 quant-ph cs.CR

classification quant-phcs.CR
keywords quantummachinelearningransomwaredetectionvariationalclassifierprincipalcomponentanalysisbarrenplateausinformationbottleneckzero-day
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 tries to establish that hybrid quantum classifiers are not currently viable for detecting unseen ransomware: after compressing a 1567-feature dataset with PCA, a variational quantum classifier on 4, 8, and 12 qubits reaches at best 55% recall, while a classical logistic regression reaches about 98%. It also tries to establish that the quantum model's failure is compound—aggressive dimensionality reduction discards most of the signal, and scaling the circuit introduces a trainability failure that makes performance non-monotonic in qubit count. A sympathetic reader should care because it turns the generic promise of quantum advantage into a concrete, measured performance gap and identifies where the bottleneck actually sits.

What carries the argument

The load-bearing object is the hybrid PCA-to-VQC pipeline. PCA projects the 1567-dimensional feature space down to n = 4, 8, or 12 components; a ZZFeatureMap encodes those components into n qubits with entangling interactions; a RealAmplitudes ansatz applies trainable Y-rotations and CNOT gates; and the COBYLA optimizer minimizes a mean-squared-error cost. The non-monotonic recall curve across 4, 8, and 12 qubits, together with the flat cost trace at 8 qubits, is the evidence that separates the information-bottleneck effect from the trainability effect.

What would settle it

Rerun the 8- and 12-qubit VQCs with the same 100-iteration budget as the 4-qubit model across at least 20 random seeds and record the cost-function traces; if the 8-qubit recall rises above the 12-qubit level or the cost descends, the non-monotonic trend is an artifact of training budget and seed, not a barren plateau. A stronger test directly estimates the variance of the cost gradient over random parameter initializations for each qubit count.

Watch

Extended reading notes

Core claim

On the Ransomware Combined Structural Feature Dataset, with ransomware families split between train and test to simulate zero-day threats, a hybrid pipeline that reduces 1567 features via PCA and classifies with a simulated variational quantum classifier reaches at best 55.06% recall at 12 qubits, against 97.66% for logistic regression on the full feature set. The authors further find that recall does not increase monotonically with qubit count: 4 qubits give 42.60%, 8 qubits drop to 40.00%, and 12 qubits recover to 55.06%. They interpret the 8-qubit stagnation, together with a flat cost-function trace, as evidence that the variational circuit enters a barren-plateau regime, a flat optimizat

Load-bearing premise

The claim that the 8-qubit model suffers a barren plateau rests on the assumption that its flat cost curve is caused by the optimization landscape and not by the unequal experiment settings—one random seed and fewer iterations for the larger models—so if those settings change the trend, the trainability conclusion loses its support.

Editorial extensions

If this is right

  • Hybrid PCA-plus-VQC does not currently offer a viable path to unseen ransomware detection: best recall is 55.06% versus 97.66% for classical logistic regression.
  • Scaling qubits from 4 to 8 degrades recall from 42.60% to 40.00% before 12 qubits improves to 55.06%, so adding quantum resources is not automatically beneficial.
  • Information loss via PCA caps the quantum model: 12 principal components retain only 35.50% of the original variance, so the VQC is starved of signal regardless of circuit expressiveness.
  • Training dynamics suggest the optimization landscape, not just data compression, is the limiting factor, which redirects future work toward trainability-preserving circuit designs.
  • Progress requires co-designing information-preserving encodings and trainable circuits rather than isolated hardware improvements.

Reading between the lines

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

  • The paper does not quantitatively separate the two failure modes, but its own logic implies a test: keep the 12-qubit input (35.5% variance) while reducing circuit difficulty—fewer parameters or a shallower ansatz—to see whether recall tracks information content or circuit size.
  • If the 8-qubit collapse is a genuine barren plateau rather than a training-budget artifact, a 16-qubit run with the same optimizer should show another plateau or collapse despite retaining more PCA variance; this is a direct, untested corollary of the trainability claim.
  • The authors suggest autoencoders as better compressors; a consequence of their dual-challenge framing is that a nonlinear compressor will only help if the ansatz and optimizer are redesigned together, since the 8-qubit failure already occurs before the information ceiling is reached.
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.

Referee Report

3 major / 4 minor

Summary. The paper reports an empirical comparison of classical machine learning models and hybrid PCA+VQC classifiers for unseen ransomware detection, using a public dataset with disjoint ransomware families. Classical baselines (especially logistic regression, 97.66% recall) are found to be far superior to the best VQC (55.06% recall at 12 qubits). The authors observe a non-monotonic trend in VQC recall versus qubit count (42.60% at 4 qubits, 40.00% at 8 qubits, 55.06% at 12 qubits) and interpret this as evidence of a severe trainability issue, specifically a barren plateau, compounded by an information bottleneck from PCA compression.

Significance. If the non-monotonic trend and the barren-plateau interpretation were robust, the paper would make a useful empirical caution about the practical limits of hybrid quantum classifiers in cybersecurity. The paper has strengths: it uses a public dataset, provides code, and explicitly compares against strong classical baselines. However, the central empirical claim rests on a single seed, unequal training budgets across configurations, and a qualitative diagnosis of landscape flatness. These issues currently undermine the paper's main conclusion, so the significance of the result is contingent on additional experimental evidence.

major comments (3)
  1. [§4.1] The non-monotonic claim rests on a comparison across three configurations that differ in two confounded ways: (i) a single global seed (42) controls all stochastic processes, and (ii) the 4-qubit VQC receives 100 COBYLA iterations while the 8- and 12-qubit models receive 80. The defining dip (recall 42.60% → 40.00%) is between 4 and 8 qubits, exactly the comparison where the iteration budget differs. Without multiple seeds and matched training effort, the observed trend is not established as a property of the quantum classifier.
  2. [§4.4] The barren plateau diagnosis is inferred from cost function values described qualitatively as 'flat' or 'stagnant,' but no plot, quantitative convergence metric, or gradient-variance data is provided. Cost stagnation under COBYLA with a single initialization could equally reflect optimizer failure, finite iteration budget, or an unlucky starting point. Please report measured gradient variance (e.g., via the parameter-shift rule) or at least cost trajectories across seeds, and relate the observed stagnation to a quantitative flatness criterion.
  3. [§5.2] The information bottleneck is asserted as a 'hard ceiling' from the cumulative explained variance (Table 1: 35.50% at 12 qubits), but variance retention is not necessarily the same as discriminative information. A necessary control is a classical model (e.g., logistic regression) trained on the same PCA-reduced features. Without this control, the performance gap between classical and VQC models cannot be cleanly attributed to the quantum classifier itself versus the compression step.
minor comments (4)
  1. [§3.4.2] The number of layers in the RealAmplitudes ansatz is not specified; Algorithm 1 line 15 and Eq. (6) refer to U(θ) without defining circuit depth. The code is available, but specifying these details would improve reproducibility.
  2. [Table 1] Minor typographical issues: 'V ariance' in Table 1, 'V ariational' in the Section 3.4.2 header, and 'T rainability' in the Section 5.2 header.
  3. [References] Reference [17] is a superconducting qubit measurement paper and does not support the claim that global observables lead to flatter landscapes; consider citing the relevant barren-plateau literature for this point.
  4. [Figure 3] With a single point per configuration and no error bars, the non-monotonic shape is visually underdetermined. Adding repeated-run distributions or error bars would help the reader assess the trend.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: empirical measurements; only a minor non-load-bearing self-citation.

full rationale

The paper is an empirical measurement study rather than a derivation chain. The central claims — that the best VQC recall (55.06%) is far below the classical baseline (97.66%), that recall is non-monotonic in qubit count (42.60% → 40.00% → 55.06%), and that PCA retains only 19.14–35.50% of the variance — are direct reported measurements on the held-out test set, not quantities defined in terms of one another. No fitted parameter is renamed as a prediction: the VQC results are test-set measurements after COBYLA training, and the PCA variance numbers are properties of the dimensionality-reduction projection, not of the VQC outcome. The non-monotonic trend is an observed result, not an input assumed by construction. The barren-plateau interpretation in §4.4 is an inference from cost-function stagnation; it is under-supported as a causal diagnosis (no gradient variance is reported, the 4-qubit model received 100 COBYLA iterations while the 8- and 12-qubit models received 80, and a single random seed was used), but that is a correctness/robustness concern, not circularity, because the diagnosis is not used to define the observed trend. The only self-citation is [18] (Do & Le), cited in §3 for general methodology framing; it is not load-bearing for the empirical results. No circular step can be exhibited with a specific Eq.-to-Eq. reduction or fitted-input-called-prediction pattern, so the honest finding is no significant circularity, with a minor non-load-bearing self-citation accounting for the score of 1.

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

The paper introduces no new theoretical entities. The central load-bearing parameters are experimental choices (iteration counts, circuit depth) that are not fully specified and may affect the key trend. The interpretation relies on domain assumptions about PCA variance and barren plateau applicability.

free parameters (3)
  • VQC training iterations = 100 for 4 qubits, 80 for 8 and 12 qubits
    Chosen by the authors to manage runtime; unequal across configurations, confounding the non-monotonic comparison.
  • Ansatz and feature map repetition counts = Not specified (Qiskit defaults likely)
    The depth of RealAmplitudes and ZZFeatureMap affects expressivity and trainability but is not reported in the paper.
  • Global random seed = 42
    Single seed with no repeated runs; results may be sensitive to initialization and data splits.
assumptions (3)
  • domain assumption Cumulative explained variance of PCA is a valid proxy for retained discriminative information
    Invoked in Section 3.4.1 and Table 1 when interpreting PCA variance as the information bottleneck; variance preservation does not necessarily equal classification-relevant information.
  • domain assumption Barren plateau theory applies at 8 and 12 qubits for this circuit
    Section 4.4 interprets flat cost curves as a barren plateau without measuring gradient variance; the crossover scale for barren plateaus is typically larger, and the claim is inferred.
  • domain assumption The dataset's train/test family separation is a realistic zero-day scenario
    Section 3.1 relies on the disjoint ransomware families to simulate unseen threats; this is a dataset design choice inherited from the reference.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Non-Monotonic Relationship: An Empirical Analysis of Hybrid Quantum Classifiers for Unseen Ransomware Detection." pith.science (2026). https://pith.science/paper/7ROBMYUE

@misc{pith2026250907924,
  author       = {Pith},
  title        = {Pith review of: A Non-Monotonic Relationship: An Empirical Analysis of Hybrid Quantum Classifiers for Unseen Ransomware Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7ROBMYUE}},
  note         = {Machine review of arXiv:2509.07924}
}
read the original abstract

Detecting unseen ransomware is a critical cybersecurity challenge where classical machine learning often fails. While Quantum Machine Learning (QML) presents a potential alternative, its application is hindered by the dimensionality gap between classical data and quantum hardware. This paper empirically investigates a hybrid framework using a Variational Quantum Classifier (VQC) interfaced with a high-dimensional dataset via Principal Component Analysis (PCA). Our analysis reveals a dual challenge for practical QML. A significant information bottleneck was evident, as even the best performing 12-qubit VQC fell short of the classical baselines 97.7\% recall. Furthermore, a non-monotonic performance trend, where performance degraded when scaling from 4 to 8 qubits before improving at 12 qubits suggests a severe trainability issue. These findings highlight that unlocking QMLs potential requires co-developing more efficient data compression techniques and robust quantum optimization strategies.

Figures

Figures reproduced from arXiv: 2509.07924 by the authors.

Figure 1
Figure 1. Experimental framework for comparing classical and hybrid quantum-classical models. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. ROC curves for classical models, demonstrating their high discriminative power on [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. A non-monotonic relationship between VQC recall and the number of qubits. Per [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

20 extracted references · 18 canonical work pages

  1. [18]

    Challenges in Applying Variational Quantum Algorithms to Dynamic Satellite Network Routing

    P. H. Do and T. D. Le, “Challenges in applying variational quantum algorithms to dynamic satellite network routing,”arXiv preprint arXiv:2508.04288, 2025. [Online]. Available: https://arxiv.org/abs/2508.04288

  2. [1]

    Machine learning-based intrusion detection for zero-day ransomware in unseen data,

    Y. Brinkley, D. Thompson, and N. Simmons, “Machine learning-based intrusion detection for zero-day ransomware in unseen data,”Authorea, 2024

  3. [2]

    Quantum machine learning: A tutorial,

    J. D. Mart ´ ın-Guerrero and L. Lamata, “Quantum machine learning: A tutorial,”Neuro- computing, vol. 470, pp. 457–461, 2022

  4. [3]

    Noisy intermediate-scale quantum algorithms,

    K. Bharti, A. Cervera-Lierta, T. H. Kyaw, T. Haug, S. Alperin-Lea, A. Anand, M. Deg- roote, H. Heimonen, J. S. Kottmann, T. Menkeet al., “Noisy intermediate-scale quantum algorithms,”Reviews of Modern Physics, vol. 94, no. 1, p. 015004, 2022

  5. [4]

    Barren plateaus in quantum neural network training landscapes,

    J. R. McClean, S. Boixo, V. N. Smelyanskiy, R. Babbush, and H. Neven, “Barren plateaus in quantum neural network training landscapes,”Nature communications, vol. 9, no. 1, p. 4812, 2018

  6. [5]

    Quantum enhanced machine learning for unobtrusive stress mon- itoring,

    A. Padha and A. Sahoo, “Quantum enhanced machine learning for unobtrusive stress mon- itoring,” inProceedings of the 2022 Fourteenth International Conference on Contemporary Computing, 2022, pp. 476–483

  7. [6]

    The role of machine learning in cybersecurity,

    G. Apruzzese, P. Laskov, E. Montes de Oca, W. Mallouli, L. Brdalo Rapa, A. V. Gram- matopoulos, and F. Di Franco, “The role of machine learning in cybersecurity,”Digital Threats: Research and Practice, vol. 4, no. 1, pp. 1–38, 2023

  8. [7]

    Recent advances for quantum classifiers,

    W. Li and D.-L. Deng, “Recent advances for quantum classifiers,”Science China Physics, Mechanics & Astronomy, vol. 65, no. 2, p. 220301, 2022. 12

Show all 20 references
  1. [8]

    Pe parser: A python package for portable executable files processing,

    D. Gibert, “Pe parser: A python package for portable executable files processing,”Software Impacts, vol. 13, p. 100365, 2022

  2. [9]

    Assessing the predictive capability of ensemble tree methods for landslide susceptibility mapping using xgboost, gradient boosting machine, and random forest,

    E. K. Sahin, “Assessing the predictive capability of ensemble tree methods for landslide susceptibility mapping using xgboost, gradient boosting machine, and random forest,”SN Applied Sciences, vol. 2, no. 7, p. 1308, 2020

  3. [10]

    Different obfuscation techniques for code protection,

    C. K. Behera and D. L. Bhaskari, “Different obfuscation techniques for code protection,” Procedia Computer Science, vol. 70, pp. 757–763, 2015

  4. [11]

    Expressibility and entangling capability of parameterized quantum circuits for hybrid quantum-classical algorithms,

    S. Sim, P. D. Johnson, and A. Aspuru-Guzik, “Expressibility and entangling capability of parameterized quantum circuits for hybrid quantum-classical algorithms,”Advanced Quantum Technologies, vol. 2, no. 12, p. 1900070, 2019

  5. [12]

    Encoding classical data into a quantum computer,

    K. Ghosh, “Encoding classical data into a quantum computer,”arXiv preprint arXiv:2107.09155, 2021

  6. [13]

    Supervised learning with quantum-enhanced feature spaces,

    V. Havl ´ ıˇ cek, A. D. C´ orcoles, K. Temme, A. W. Harrow, A. Kandala, J. M. Chow, and J. M. Gambetta, “Supervised learning with quantum-enhanced feature spaces,”Nature, vol. 567, no. 7747, pp. 209–212, 2019

  7. [14]

    Modeling feature maps for quantum machine learning,

    N. Singh and S. R. Pokhrel, “Modeling feature maps for quantum machine learning,”arXiv preprint arXiv:2501.08205, 2025

  8. [15]

    Classification of medical image modeling methods: A review,

    Z. Amini and H. Rabbani, “Classification of medical image modeling methods: A review,” Current Medical Imaging Reviews, vol. 12, no. 2, pp. 130–148, 2016

  9. [16]

    A comparison of various classical optimizers for a variational quantum linear solver,

    A. Pellow-Jarman, I. Sinayskiy, A. Pillay, and F. Petruccione, “A comparison of various classical optimizers for a variational quantum linear solver,”Quantum Information Pro- cessing, vol. 20, no. 6, p. 202, 2021

  10. [17]

    Simultaneous state measurement of coupled josephson phase qubits,

    R. McDermott, R. Simmonds, M. Steffen, K. B. Cooper, K. Cicak, K. Osborn, S. Oh, D. P. Pappas, and J. M. Martinis, “Simultaneous state measurement of coupled josephson phase qubits,”Science, vol. 307, no. 5713, pp. 1299–1302, 2005

  11. [19]

    Ransomware combined structural feature dataset,

    C. Moreira, D. Moreira, and C. Sales Jr., “Ransomware combined structural feature dataset,” 2024. [Online]. Available: https://doi.org/10.17632/yzhcvn7sj5.2

  12. [20]

    Variational quantum classifier for binary classification: Real vs synthetic dataset,

    D. Maheshwari, D. Sierra-Sosa, and B. Garcia-Zapirain, “Variational quantum classifier for binary classification: Real vs synthetic dataset,”IEEE access, vol. 10, pp. 3705–3715, 2021. 13

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.