Pith. sign in

REVIEW 2 major objections 4 minor 37 references

Noisy-QSMOTE: Robustness Analysis of Quantum SMOTE under Quantum-Inspired Noise for Condition Monitoring and Fault Classification in Industrial and Energy Systems

T0 review · 2 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read QSMOTE is claimed to fix class imbalance in industrial fault data, boosting random forests and SVMs by up to 170% on engine-failure data and to over 0.99 accuracy on industrial-fault data, while linear and naive Bayes models often degrade.

desk verdict A useful empirical benchmark idea undercut by a non-integer qubit count in Algorithm 1 that makes the core QSMOTE results unreproducible without an undocumented workaround. read the letter →

arxiv 2601.11423 v2 pith:UPQ6GIVW submitted 2026-01-16 quant-ph

Amit S. Patel , Himanshukumar R. Patel , Bikash K. Behera This is my paper · ORCID
classification quant-ph
keywords quantumSMOTEclassimbalanceoversamplingfaultdiagnosisconditionmonitoringnoiserobustnessmachinelearning
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

QSMOTE is claimed to solve class imbalance in industrial and energy fault-detection datasets by generating synthetic minority samples whose step length is governed by a quantum-inspired similarity angle. Across four datasets, the paper reports consistent large gains for non-linear classifiers—random forests, decision trees, and SVMs—with relative improvements up to about 170% on the Engine Failure Detection Dataset and near-perfect accuracy (over 0.99) on the Industrial Fault Detection Dataset. The same results show that linear and probabilistic models, particularly logistic regression and naive Bayes, often lose accuracy after oversampling, which the paper attributes to interpolation-induced boundary distortion in overlapping feature spaces. In a second stage, the paper injects six quantum-noise channels into the similarity-estimation step and claims that ensemble and margin-based models retain most of their accuracy even at maximum noise, while naive Bayes and decision trees degrade substantially, with depolarizing noise the most disruptive. If these claims hold, they give practitioners a clear baseline: use QSMOTE with an ensemble or SVM for imbalanced, noisy industrial monitoring, and avoid simple linear or probabilistic models.

What carries the argument

The mechanism is the compact swap test used to estimate the quantum overlap between a minority sample and its cluster centroid. The sample and centroid are amplitude-encoded as quantum states; the circuit measures the ancilla in a way that encodes their inner product, which is converted to a 'quantum-inspired angle' by arccos after clipping to [-1,1]. A uniform step length is then sampled from [0, angle/split_factor], and the synthetic sample is placed at x + step * (centroid - x)/||centroid - x||. The angle therefore directly controls how far synthetic points move: a poorly aligned sample (large angle) moves farther toward its centroid, while an aligned sample stays close to x. The noise an

What would settle it

Inspect the experiment code or rerun the pipeline for a dataset with a non-power-of-two feature count, such as CWRUBD's nine time-domain features. If Algorithm 1 is applied as written, constructing a circuit with n = log2(9) + 3 qubits (a fractional number) should fail in any standard quantum circuit simulator; if the code silently pads, truncates, or switches to a classical inner product, then the QSMOTE samples were not generated by the amplitude-encoded compact swap test, and the reported accuracy gains are attributable to whatever substitute was actually used. A reader could also compare t

Watch

Extended reading notes

Core claim

The paper's central claim is that QSMOTE—which generates a synthetic minority sample by moving the point toward its K-means cluster centroid by a distance drawn uniformly from zero to a 'quantum-inspired angle' between the point and the centroid divided by a split factor—substantially improves non-linear classifiers on imbalanced industrial fault data. The reported effect is model- and dataset-dependent: on EFDD, random forest accuracy rises from 0.4923 to 0.9126 and decision tree from 0.2979 to 0.8228; on IFDD, random forest reaches 0.9919 and decision tree 0.9006; on SPID, SVM rises from 0.8057 to 0.8893. Logistic regression and naive Bayes often fall, with LR on EFDD dropping from 0.5337

Load-bearing premise

The paper's load-bearing premise is that the compact swap test in Algorithm 1 can actually be executed on the datasets' feature vectors; the qubit count is set to log2(len(c)) + 3, which is not a whole number for feature vectors whose length is not a power of two (CWRUBD has nine features), and no padding or alternative amplitude-encoding step is described—if the circuit cannot be constructed, the reported synthetic samples and all downstream accuracy figures are not produced

Editorial extensions

If this is right

  • For imbalanced industrial and energy monitoring data, QSMOTE plus a random forest or SVM is a strong combination: reported accuracies exceed 0.99 on IFDD and 0.98 on CWRUBD.
  • Model-family guidance emerges: non-linear, ensemble, and margin-based classifiers benefit from QSMOTE; logistic regression and naive Bayes often degrade, so resampling choices must be model-aware.
  • The noise results give a simple selection rule for noisy environments: random forests retain near-baseline accuracy at maximum noise, while naive Bayes and logistic regression should be avoided when quantum-inspired perturbations are possible.
  • Depolarizing noise being the most disruptive channel and phase-damping the least suggests where to direct noise-mitigation effort in quantum-inspired resampling pipelines.
  • The reported 170% relative gain on EFDD provides a concrete performance target for validating future balancing methods against this baseline.

Reading between the lines

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

  • A classical variant may capture most of the effect: if the gain comes mainly from the angle-scaled step toward cluster centroids, replacing the swap-test angle with a classical cosine similarity between x and its centroid would be a cheap, falsifiable proxy to test.
  • The optional convex clamp mentioned in the paper is not used in Algorithm 2; enabling it would force synthetic points onto the segment from x to its centroid and is a natural extension to see whether the reported LR/NB degradation shrinks.
  • Noise injected into the similarity estimator is a distinct failure mode from noise on raw sensor features; comparing the two would show whether quantum-inspired perturbation models are the right stress test for real industrial sensor corruption.
  • Because the swap-test estimate is based on a fixed shot count (1000), sampling noise in the angle itself is a source of stochasticity; increasing shots should tighten angle estimates and could reduce variance in the reported gains, a testable prediction.
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

2 major / 4 minor

Summary. The paper presents a benchmark study of Quantum SMOTE (QSMOTE) implemented via a compact swap test, applied to four industrial condition-monitoring datasets (SPID, CWRUBD, EFDD, IFDD) and evaluated with five classical classifiers (LR, RF, SVM, DT, NB) before and after oversampling. The authors also report a robustness analysis under six quantum-inspired noise channels and claim that QSMOTE substantially improves non-linear classifiers, with gains up to roughly 170% on EFDD and near-perfect accuracy on IFDD. The manuscript's main contributions are framed as the first joint evaluation of imbalance mitigation and quantum-noise robustness for industrial fault classification.

Significance. If correct, the paper would provide a useful empirical reference for practitioners combining quantum-inspired oversampling with classical classifiers in industrial diagnostics. The strengths are the breadth of datasets and classifiers, the internally consistent before/after tables in Tables 1–4, and the explicit statement of the main free parameters (κ, K, noise probability grid). I agree with the reader that the tables are internally consistent as comparisons. However, two load-bearing details are missing: the circuit in Algorithm 1 cannot be constructed as written for the datasets used, and the noise-injection protocol is not operationally defined. These gaps prevent the reported results from being reproduced from the described method, so the central claims are not yet established.

major comments (2)
  1. [Section 3.2, Eqs. (7)–(9)] The derivation of the overlap estimator is asserted rather than shown. The text writes P0 as a single norm-squared expression, then jumps to Eq. (8), 'after simplifying,' with no intermediate algebra. The notation also mixes state vectors |C⟩,|M⟩ with the scalar normalization constants C and M; Eq. (3) defines |ψ⟩ with states |C⟩,|M⟩ but Eq. (4) and onward appear to treat C and M as scalars. Since the compact swap test is the foundation of the quantum-inspired angle α(x,c) used in Algorithm 2, a correct and complete derivation — or an explicit citation to a source where this relation is proved — is required before the QSMOTE geometry can be considered well-defined.
  2. [Section 5, conclusion] The conclusion states 'DT on IFDD improving from 0.7393 to 0.9006,' but Table 4 reports IFDD DT accuracy before QSMOTE as 0.4685 and after as 0.9006. The value 0.7393 does not appear in Table 4 and appears to be taken from an unrelated metric or table. This inconsistency suggests a copy‑paste error in a central summary statement and should be corrected.
minor comments (4)
  1. [Abstract] The abstract refers to 'Linear Regression (LR)' while the method section and tables use logistic regression (LR). The model is logistic regression, so the terminology should be corrected.
  2. [Algorithm 1, line 8] The single-qubit rotation u(θ, π, 0) is not defined. If this is a Qiskit-style U3 gate, it should be stated explicitly, or the parameters should be named consistently with the circuit diagram.
  3. [Section 4.1] The EFDD and IFDD descriptions say each has 1,000 samples but do not state the number of features or the exact class counts before pre-processing. Since Algorithm 1's qubit count depends on the feature dimension, these counts should be reported.
  4. [Appendix A] Tables A1–A16 are not cited anywhere in the main text, and the appendix is mentioned only as 'Supplementary information.' These additional results should either be integrated into the main narrative or removed, and the caption of Figs. A1/A2 has mismatched subfigure labels (the caption lists (b)–(f) but the figure appears to show different panels).

Circularity Check

1 steps flagged · score 2.0 of 10

Only circular element: the claim that QSMOTE balances classes is true by Algorithm 2's definition; the classifier and noise-robustness results are independent empirical benchmarks.

  1. self definitional [Algorithm 2 (Section 3.3), lines 8-9 and 32; Section 4.6, Fig. 3 discussion]
    "8:foreach minority class arraydo 9:Compute number of new samples neededneed←majority count−len(minority class) ... The minority classes are synthetically augmented to obtain a balanced representation in comparison to the majority classes. Comparing the two sets of subfigures demonstrates that QSMOTE effectively equalizes class frequencies across all datasets."

    Algorithm 2 defines need as majority count minus minority count and appends exactly that many synthetic samples, so the post-QSMOTE class counts are equal by construction. Section 4.6 presents Fig. 3 as though it empirically demonstrates equalization, but the 'result' is just the algorithm's definition restated. This is a self-definitional observation rather than a prediction with independent content. It does not extend to the classifier accuracy gains or the noise-robustness hierarchy, which are genuine external benchmarks.

full rationale

The paper's main deliverables—accuracy/precision/recall/F1 before and after QSMOTE, the per-model gains, and the six-channel noise robustness trends—are external empirical benchmarks computed from datasets and classifiers, not derivations from fitted parameters. No step reduces a predicted classifier accuracy to a fitted input: QSMOTE's angle and step length are computed from the same features, but the downstream classifier outcomes are not constrained by construction (indeed LR and NB sometimes degrade, so the comparison has empirical content). The reliance on the authors' own QSMOTE paper [8] is real but not load-bearing here: Algorithm 2 is fully specified in the manuscript and the present claims are validated by the reported experiments rather than by [8]'s authority. The only by-construction item is the observation that QSMOTE equalizes class frequencies: Algorithm 2 explicitly sets need = majority count − minority count and appends that many synthetic samples, so Fig. 3's 'demonstration' restates the algorithm's definition. This is a minor self-definitional step and does not affect the independent classifier benchmarks. The non-integer qubit count in Algorithm 1 (n = log2(len(c)) + 3) is a reproducibility/correctness concern, not a circularity.

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

The central empirical claims rest on the QSMOTE implementation. The paper supplies free parameters (split factor κ, cluster count K, noise grid) without exact values. It also assumes the compact swap test is valid for arbitrary feature dimensions and that the six quantum noise channels, described only as Kraus operators, can be meaningfully inserted into the swap-test circuit. No invented physical entities are introduced.

free parameters (3)
  • split factor κ
    User-chosen step-length divisor in Eq. (16); no value is reported. It directly scales synthetic sample displacement, so all results depend on it.
  • number of clusters K
    K-means cluster count for each dataset; not reported. Determines centroids and therefore the direction of synthetic samples.
  • noise probability grid
    The x-axis of Figs. 4–7 is 'noise probability', but exact grid values and the mapping of Kraus operator probabilities to circuit error are not specified.
assumptions (4)
  • domain assumption K-means centroids capture the local geometry of minority classes in standardized feature space
    QSMOTE moves synthetic samples toward the assigned centroid; if the clustering is not meaningful, the oversampled data does not respect the manifold (Section 3.1, Eq. 10).
  • ad hoc to paper Eq. (8) gives the overlap between amplitude-encoded vectors without error
    The derivation in Section 3.2 ends with Eq. (8) after 'simplifying'; no proof or external reference is given, and the circuit definition in Algorithm 1 is incomplete for non-power-of-two dimensions.
  • domain assumption Quantum noise channels can be applied to the compact swap test such that the noisy angle remains a plausible similarity measure for synthetic sample generation
    Section 4.5 lists Kraus operators but does not specify how they are inserted into the circuit or how the resulting overlap error translates into Eq. (16) step sizes.
  • domain assumption Standardization makes the dimensionless angle α map to meaningful feature-space step lengths
    Stated as 'Scale compatibility' in Section 3.3; required for Eqs. (16)-(17) but not validated on the four datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Noisy-QSMOTE: Robustness Analysis of Quantum SMOTE under Quantum-Inspired Noise for Condition Monitoring and Fault Classification in Industrial and Energy Systems." pith.science (2026). https://pith.science/paper/UPQ6GIVW

@misc{pith2026260111423,
  author       = {Pith},
  title        = {Pith review of: Noisy-QSMOTE: Robustness Analysis of Quantum SMOTE under Quantum-Inspired Noise for Condition Monitoring and Fault Classification in Industrial and Energy Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UPQ6GIVW}},
  note         = {Machine review of arXiv:2601.11423}
}
read the original abstract

Imbalanced datasets remain a major challenge in industrial condition monitoring and fault diagnosis, often causing machine-learning models to favor majority classes while underrepresenting minority fault conditions. This work investigates the Quantum Synthetic Minority Oversampling Technique (QSMOTE) through three stages: (i) baseline evaluation on the original imbalanced datasets, (ii) assessment after QSMOTE-based balancing, and (iii) analysis of QSMOTE under quantum-inspired perturbations. Unlike conventional robustness studies, the considered noise channels are injected directly into the compact-swap-test-based similarity estimation process used during synthetic sample generation, influencing overlap estimation, angle computation, and the generated minority samples. Experiments are conducted on four multi-class datasets: the Solar Panel Image Dataset (SPID), the CWRU Bearing Dataset (CWRUBD), the Engine Failure Detection Dataset (EFDD), and the Industrial Fault Detection Dataset (IFDD). Performance is evaluated using Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), Logistic Regression (LR), and Naive Bayes (NB) classifiers. The results show that QSMOTE effectively reduces class imbalance and substantially improves the performance of non-linear classifiers, with gains of up to 170% on EFDD and accuracies exceeding 0.99 on IFDD. Further analysis under bit-flip, phase-flip, bit-phase-flip, depolarizing, amplitude damping, and phase damping channels demonstrates how perturbations introduced during similarity estimation propagate through synthetic sample generation and influence downstream classification performance. The proposed framework provides a practical approach for studying both imbalance mitigation and noisy quantum-inspired oversampling in industrial and energy-system applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 7 canonical work pages

  1. [1]

    Procedia CIRP38, 3–7 (2015) https://doi.org/10.1016/j.procir.2015.08.026

    Lee, J., Ardakani, H.D., Yang, S., Bagheri, B.: Industrial big data analytics and cyber-physical systems for future maintenance & service innovation. Procedia CIRP38, 3–7 (2015) https://doi.org/10.1016/j.procir.2015.08.026

    work page doi:10.1016/j.procir.2015.08.026 2015

  2. [2]

    SSRG International Journal of Electrical and Electronics Engineering11(9), 294–315 (2024) https://doi.org/10.14445/23488379/IJEEE-V11I9P127

    Vijayalakshmi, K., Rajakannu, A., Kamarudden, M., Ramachandran, K.P., Sri Rajkavin, A.V.: Intelligent fault diagnosis of rotating machinery using deep learning algorithms: A comparative analysis of mlp, cnn, rnn, and lstm. SSRG International Journal of Electrical and Electronics Engineering11(9), 294–315 (2024) https://doi.org/10.14445/23488379/IJEEE-V11I9P127

    work page doi:10.14445/23488379/ijeee-v11i9p127 2024

  3. [3]

    Journal of Artificial Intelligence Research16, 321–357 (2002)

    Chawla, N.V., Bowyer, K.W., Hall, L.O., Kegelmeyer, W.P.: Smote: Synthetic minority over-sampling technique. Journal of Artificial Intelligence Research16, 321–357 (2002)

    2002

  4. [4]

    IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (2009) https://doi.org/10

    He, H., Garcia, E.A.: Learning from imbalanced data. IEEE Transactions on Knowledge and Data Engineering21(9), 1263–1284 (2009) https://doi.org/10. 1109/TKDE.2008.239

    2009

  5. [5]

    AIP Advances15(2), 020702 (2025) https://doi.org/10.1063/5

    Zhao, J., Wang, W., Huang, J., Ma, X.: A comprehensive review of deep learning-based fault diagnosis approaches for rolling bearings: Advancements and challenges. AIP Advances15(2), 020702 (2025) https://doi.org/10.1063/5. 0255451

    doi:10.1063/5 2025

  6. [6]

    Engineering Applications of Artificial Intelligence163(Part 4), 113068 (2026) https://doi.org/10.1016/j.engappai.2025.113068

    Bao, Z., Liu, C., Yang, H., Zhang, J., Li, Y.: From theory to industry: A sur- vey of deep learning-enabled bearing fault diagnosis in complex environments. Engineering Applications of Artificial Intelligence163(Part 4), 113068 (2026) https://doi.org/10.1016/j.engappai.2025.113068

    arXiv 2026

  7. [7]

    Quantum2, 79 (2018) https://doi.org/10.22331/q-2018-08-06-79

    Preskill, J.: Quantum Computing in the NISQ era and beyond. Quantum2, 79 (2018) https://doi.org/10.22331/q-2018-08-06-79

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

  8. [8]

    Quantum Machine Intelligence7(1), 38 (2025) https://doi.org/10.1007/s42484-025-00248-6

    Mohanty, N., Behera, B.K., Ferrie, C., Dash, P.: A quantum approach to synthetic 41 minority oversampling technique (smote). Quantum Machine Intelligence7(1), 38 (2025) https://doi.org/10.1007/s42484-025-00248-6

    work page doi:10.1007/s42484-025-00248-6 2025

Show all 37 references
  1. [9]

    Contemporary Physics56(2), 172–185 (2015) https://doi.org/10.1080/ 00107514.2014.964942

    Schuld, M., Sinayskiy, I., Petruccione, F.: An introduction to quantum machine learning. Contemporary Physics56(2), 172–185 (2015) https://doi.org/10.1080/ 00107514.2014.964942

    2015

  2. [10]

    Nature549(7671), 195–202 (2017) https://doi.org/ 10.1038/nature23474

    Biamonte, J., Wittek, P., Pancotti, N., Rebentrost, P., Wiebe, N., Lloyd, S.: Quantum machine learning. Nature549(7671), 195–202 (2017) https://doi.org/ 10.1038/nature23474

    2017 doi

  3. [11]

    Physical Review A98(1), 012324 (2018) https://doi.org/10.1103/PhysRevA.98

    Dallaire-Demers, P.-L., Killoran, N.: Quantum generative adversarial networks. Physical Review A98(1), 012324 (2018) https://doi.org/10.1103/PhysRevA.98. 012324

    2018 doi

  4. [12]

    Physical Review A96(6), 062327 (2017) https://doi.org/ 10.1103/PhysRevA.96.062327

    Kieferov´ a, M., Wiebe, N.: Tomography and generative training with quantum boltzmann machines. Physical Review A96(6), 062327 (2017) https://doi.org/ 10.1103/PhysRevA.96.062327

    2017 doi

  5. [13]

    Measurement242(Part D), 116216 (2025) https://doi

    Vashishtha, G., Chauhan, S., Sehri, M., Zimroz, R., Dumond, P., Kumar, R., Gupta, M.K.: A roadmap to fault diagnosis of industrial machines via machine learning: A brief review. Measurement242(Part D), 116216 (2025) https://doi. org/10.1016/j.measurement.2024.116216

    2025

  6. [14]

    BMC Bioinformatics14(1), 106 (2013) https://doi.org/10.1186/1471-2105-14-106

    Blagus, R., Lusa, L.: Smote for high-dimensional class-imbalanced data. BMC Bioinformatics14(1), 106 (2013) https://doi.org/10.1186/1471-2105-14-106

    2013 doi

  7. [15]

    Springer Series in Statistics

    Fern´ andez, A., Garc ´ ıa, S., Galar, M., Prati, R.C., Krawczyk, B., Herrera, F.: Learning from Imbalanced Data Sets. Springer Series in Statistics. Springer, Cham (2018). https://doi.org/10.1007/978-3-319-98074-4

    2018 doi

  8. [16]

    New Journal of Physics18(2), 023023 (2016) https://doi.org/10.1088/1367-2630/18/2/023023

    McClean, J.R., Romero, J., Babbush, R., Aspuru-Guzik, A.: The theory of varia- tional hybrid quantum–classical algorithms. New Journal of Physics18(2), 023023 (2016) https://doi.org/10.1088/1367-2630/18/2/023023

    2016 doi

  9. [17]

    Quantum Science and Technology4(4), 043001 (2019) https://doi.org/10.1088/2058-9565/ab4eb5

    Benedetti, M., Lloyd, E., Sack, S., Fiorentini, M.: Parameterized quantum cir- cuits as machine learning models. Quantum Science and Technology4(4), 043001 (2019) https://doi.org/10.1088/2058-9565/ab4eb5

    2019 doi

  10. [18]

    Proceedings of the 2005 International Conference on Advances in Intelligent Computing (ICIC’05), Volume Part I, 878–887 (2005) https://doi.org/10.1007/11538059 91

    Han, H., Wang, W.-Y., Mao, B.-H.: Borderline-smote: A new over-sampling method in imbalanced data sets learning. Proceedings of the 2005 International Conference on Advances in Intelligent Computing (ICIC’05), Volume Part I, 878–887 (2005) https://doi.org/10.1007/11538059 91

    2005 doi

  11. [19]

    IEEE International Joint Conference on Neural Networks, 1322–1328 (2008) https://doi.org/10.1109/IJCNN.2008.4633969 42

    He, H., Bai, Y., Garcia, E.A., Li, S.: Adasyn: Adaptive synthetic sampling approach for imbalanced learning. IEEE International Joint Conference on Neural Networks, 1322–1328 (2008) https://doi.org/10.1109/IJCNN.2008.4633969 42

    2008

  12. [20]

    SIGKDD Explorations 6(1), 20–29 (2004) https://doi.org/10.1145/1007730.100773

    Batista, G.E., Prati, R.C., Monard, M.C.: A study of the behavior of several methods for balancing machine learning training data. SIGKDD Explorations 6(1), 20–29 (2004) https://doi.org/10.1145/1007730.100773

    2004

  13. [21]

    Vibration6(1), 218–238 (2023) https://doi.org/10.3390/vibration6010014

    Bhuiyan, M.R., Uddin, J.: Deep transfer learning models for industrial fault diag- nosis using vibration and acoustic sensors data: A review. Vibration6(1), 218–238 (2023) https://doi.org/10.3390/vibration6010014

    2023 doi

  14. [22]

    Applied Sciences15(21), 11303 (2025) https://doi.org/10.3390/app152111303

    Wang, H., Wang, H., Tang, X.: A review of deep learning in rotating machinery fault diagnosis and its prospects for port applications. Applied Sciences15(21), 11303 (2025) https://doi.org/10.3390/app152111303

    2025 doi

  15. [23]

    2019 International Conference on Control, Automation and Diagnosis (ICCAD) (2019) https://doi.org/10.1109/ ICCAD46983.2019.9037949

    Lo, N.G., Flaus, J.-M., Adrot, O.: Review of machine learning approaches in fault diagnosis applied to iot systems. 2019 International Conference on Control, Automation and Diagnosis (ICCAD) (2019) https://doi.org/10.1109/ ICCAD46983.2019.9037949

    2019

  16. [24]

    Technical Report, IBM Research (2001)

    Rish, I.: An Empirical Study of the Naive Bayes Classifier. Technical Report, IBM Research (2001)

    2001

  17. [25]

    IEEE Transactions on Neural Networks10(5), 988–999 (1999) https://doi.org/10.1109/72.788640

    Vapnik, V.N.: An overview of statistical learning theory. IEEE Transactions on Neural Networks10(5), 988–999 (1999) https://doi.org/10.1109/72.788640

    1999 doi

  18. [26]

    Machine Learning45(1), 5–32 (2001) https://doi

    Breiman, L.: Random forests. Machine Learning45(1), 5–32 (2001) https://doi. org/10.1023/A:1010933404324

    2001 doi

  19. [27]

    In: arXiv:1312.6199 (2014)

    Szegedy, C., Zaremba, W., Sutskever, I.,et al.: Intriguing properties of neural networks. In: arXiv:1312.6199 (2014). https://doi.org/10.48550/arXiv.1312.6199

    2014 doi arXiv

  20. [28]

    Hendrycks, D., Dietterich, T.: Benchmarking neural network robustness to common corruptions and perturbations. (2019). https://doi.org/10.48550/arXiv. 1903.12261

    2019 doi arXiv

  21. [29]

    Measurement257(Part A), 118564 (2026) https://doi.org/10

    Song, Q., Sun, S., Wang, B., Song, Q., Wang, T., Jiang, H.: Noise-robust fault diagnosis network based on multiscale feature enhancement and dynamic cross- modal interaction. Measurement257(Part A), 118564 (2026) https://doi.org/10. 1016/j.measurement.2025.118564

    2026

  22. [30]

    Nielsen, M.A., Chuang, I.L.: Quantum computation and quantum information (2002) https://doi.org/10.1017/CBO9780511976667

    2002 doi

  23. [31]

    Physical Review Letters122(4), 040504 (2019) https://doi.org/10.1103/ PhysRevLett.122.040504

    Schuld, M., Killoran, N.: Quantum machine learning in feature hilbert spaces. Physical Review Letters122(4), 040504 (2019) https://doi.org/10.1103/ PhysRevLett.122.040504

    2019

  24. [32]

    arXiv preprint arXiv:2409.07632 (2024) https://doi.org/ 43 10.48550/arXiv.2409.07632 arXiv:2409.07632 [quant-ph]

    Khanal, B., Rivas, P.: Learning robust observable to address noise in quan- tum machine learning. arXiv preprint arXiv:2409.07632 (2024) https://doi.org/ 43 10.48550/arXiv.2409.07632 arXiv:2409.07632 [quant-ph]

    2024 doi arXiv

  25. [33]

    https://www.kaggle.com/datasets/ pythonafroz/solar-panel-images

    Afroz, P.: Solar Panel Images Dataset. https://www.kaggle.com/datasets/ pythonafroz/solar-panel-images. Kaggle Dataset. Accessed: 2025-09-01

    2025

  26. [34]

    https://www.kaggle.com/datasets/brjapon/ cwru-bearing-datasets

    brjapon: CWRU Bearing Datasets. https://www.kaggle.com/datasets/brjapon/ cwru-bearing-datasets. Kaggle dataset. Accessed: 2025-09-26

    2025

  27. [35]

    https://www.kaggle.com/datasets/ ziya07/engine-failure-detection-dataset

    ziya07: Engine Failure Detection Dataset. https://www.kaggle.com/datasets/ ziya07/engine-failure-detection-dataset. Kaggle dataset. Accessed: 2025-09-30

    2025

  28. [36]

    https://www.kaggle.com/ datasets/programmer3/industrial-fault-detection-dataset

    Programmer3: Industrial Fault Detection Dataset. https://www.kaggle.com/ datasets/programmer3/industrial-fault-detection-dataset. Kaggle dataset. Accessed: 2025-10-01

    2025

  29. [37]

    IEEE Internet of Things Journal 11(3), 3840–3852 (2023) https://doi.org/10.1109/JIOT.2023.3300577 44

    Satpathy, S.K., Vibhu, V., Behera, B.K., Al-Kuwari, S., Mumtaz, S., Farouk, A.: Analysis of quantum machine learning algorithms in noisy channels for classifi- cation tasks in the iot extreme environment. IEEE Internet of Things Journal 11(3), 3840–3852 (2023) https://doi.org/...

    2023

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.