pith. sign in

arxiv: 2605.16067 · v1 · pith:7XLGFUGLnew · submitted 2026-05-15 · 💻 cs.LG · stat.ML

SAFE Quantum Machine Learning with Variational Quantum Classifiers

Ying Chen , Paolo Giudici , Vasily Kolesnikov , Paolo Recchia This is my paper

Pith reviewed 2026-05-20 20:43 UTC · model grok-4.3

classification 💻 cs.LG stat.ML
keywords variational quantum classifieramplitude encodingSAFE-AI metricsrobustnessstabilityquantum machine learningsafety-critical applications
0
0 comments X

The pith

Quantum classifier matches classical accuracy with improved robustness and stability

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

The paper proposes a variational quantum classifier that processes high-dimensional deep representations via amplitude encoding, stabilized by a learnable classical pre-encoding layer. By combining normalized amplitude embeddings with bounded quantum observables, the model creates a structured and smooth hypothesis class with controlled sensitivity to input variations. Reliability is assessed using SAFE-AI metrics derived from the Cramer-von Mises divergence for consistent evaluation across accuracy, robustness, and explainability. Empirical results indicate the quantum model provides competitive predictive performance compared to classical baselines while showing a more balanced SAFE reliability profile with better robustness to noise and stability under structured feature removal.

Core claim

The proposed variational quantum classifier uses amplitude encoding on high-dimensional representations stabilized by a classical pre-encoding layer. Combining normalized amplitude embeddings with bounded quantum observables induces a structured and smooth hypothesis class with controlled sensitivity to input variations. This allows for reliable assessment via SAFE-AI metrics from the Cramer-von Mises divergence, leading to competitive performance with classical baselines and a more balanced profile in robustness and stability for safety-critical settings.

What carries the argument

Normalized amplitude embeddings combined with bounded quantum observables inducing a structured and smooth hypothesis class with controlled sensitivity to input variations

If this is right

  • The model achieves competitive predictive performance compared with strong classical baselines.
  • It exhibits a more balanced SAFE reliability profile.
  • It has improved robustness to noise.
  • It demonstrates stability under structured feature removal.

Where Pith is reading between the lines

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

  • This could make quantum models preferable in environments with high input variability or noise.
  • The metrics based on Cramer-von Mises divergence might be useful for other types of machine learning reliability assessments.
  • Hybrid quantum-classical setups like this pre-encoding layer could be explored for optimizing other safety properties.

Load-bearing premise

Combining normalized amplitude embeddings with bounded quantum observables induces a structured and smooth hypothesis class with controlled sensitivity to input variations.

What would settle it

If experiments show that the quantum model does not have improved robustness to noise or stability under structured feature removal compared to classical models, the claim of a more balanced SAFE reliability profile would be falsified.

Figures

Figures reproduced from arXiv: 2605.16067 by Paolo Giudici, Paolo Recchia, Vasily Kolesnikov, Ying Chen.

Figure 1
Figure 1. Figure 1: , the proposed quantum machine learning pipeline is depicted. . . . 512 ResNet . . . GELU Pre-VQC |ψ⟩ R(α1, β1, γ1) Z R(α2, β2, γ2) Z R(α3, β3, γ3) Z 9 qubits Amplitude Strongly Entangling PQC. VQC 3 Classes [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Quantum-based tumor classification on unseen MRI scans. Each image [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: RGA (mean across 5 folds). Degradation of ranking accuracy as a function of the fraction of removed samples. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: RGR (mean across 5 folds). Robustness to additive Gaussian noise under increasing noise standard deviation. 17 [PITH_FULL_IMAGE:figures/full_fig_p017_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Robustness curves under FGSM feature perturbations (top) and spatial [PITH_FULL_IMAGE:figures/full_fig_p018_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Example of progressive image occlusion used in the RGE analysis. The original MRI image is shown alongside increasingly occluded versions obtained by masking the most relevant regions according to Grad￾CAM–based patch ranking [PITH_FULL_IMAGE:figures/full_fig_p020_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: RGE explainability robustness curves on the MRI dataset under [PITH_FULL_IMAGE:figures/full_fig_p020_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Radar plot summarising F1-macro and SAFE-AI metrics. All metrics [PITH_FULL_IMAGE:figures/full_fig_p022_8.png] view at source ↗
read the original abstract

We propose a variational quantum classifier operating on high dimensional deep representations via amplitude encoding, stabilized by a learnable classical pre encoding layer.By combining normalized amplitude embeddings with bounded quantum observables, the resulting model induces a structured and smooth hypothesis class with controlled sensitivity to input variations. Model reliability is assessed using SAFE-AI metrics derived from the Cramer von Mises divergence, enabling consistent evaluation across accuracy, robustness, and explainability dimensions. Empirical results show that the proposed quantum model provides competitive predictive performance compared with strong classical baselines while exhibiting a more balanced SAFE reliability profile, with improved robustness to noise and stability under structured feature removal. These findings suggest that variational quantum circuits offer a principled mechanism for stability oriented SAFE learning in safety critical settings.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The paper proposes a variational quantum classifier that processes high-dimensional deep representations via amplitude encoding, stabilized by a learnable classical pre-encoding layer. It claims that normalized amplitude embeddings combined with bounded quantum observables induce a structured, smooth hypothesis class with controlled sensitivity to input variations. Model reliability is evaluated using SAFE-AI metrics derived from the Cramér-von Mises divergence across accuracy, robustness, and explainability. Empirical results are presented claiming competitive predictive performance relative to strong classical baselines, along with a more balanced SAFE reliability profile featuring improved robustness to noise and stability under structured feature removal.

Significance. If the central mechanistic claim holds, the work could contribute to stability-oriented quantum machine learning for safety-critical applications by linking variational circuits to controlled sensitivity and SAFE evaluation. The use of Cramér-von Mises-based metrics for multi-dimensional reliability assessment is a potentially useful framing, but the significance is limited by the absence of an explicit sensitivity bound or separation of quantum versus classical contributions.

major comments (2)
  1. [Abstract] Abstract and theoretical motivation section: The assertion that normalized amplitude embeddings with bounded quantum observables induce 'controlled sensitivity to input variations' is load-bearing for the explanation of the improved SAFE profile, yet no derivation or explicit bound (e.g., end-to-end Lipschitz constant through the encoding map, variational circuit, and observable) is supplied. Without this, it remains possible that observed robustness arises primarily from the classical pre-encoding layer.
  2. [Empirical Evaluation] Empirical results section: The claim of competitive performance and improved robustness/stability under noise and feature removal is presented without reported dataset sizes, number of independent runs, error bars, or statistical tests comparing the quantum model against baselines, undermining assessment of whether the data support the stated superiority in the SAFE dimensions.
minor comments (2)
  1. [Method] Clarify how the learnable classical pre-encoding layer parameters are optimized jointly with the variational circuit and whether any regularization is applied to isolate its effect on the hypothesis class.
  2. [SAFE Metrics] The notation for the SAFE metrics should explicitly state the precise application of the Cramér-von Mises divergence to each reliability dimension and how thresholds or normalizations are chosen.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help clarify the presentation of our theoretical claims and empirical results. We address each major comment below and describe the revisions we will incorporate.

read point-by-point responses

  1. Referee: [Abstract] Abstract and theoretical motivation section: The assertion that normalized amplitude embeddings with bounded quantum observables induce 'controlled sensitivity to input variations' is load-bearing for the explanation of the improved SAFE profile, yet no derivation or explicit bound (e.g., end-to-end Lipschitz constant through the encoding map, variational circuit, and observable) is supplied. Without this, it remains possible that observed robustness arises primarily from the classical pre-encoding layer.

    Authors: We agree that an explicit end-to-end sensitivity bound would strengthen the mechanistic explanation. The manuscript motivates controlled sensitivity from the unit-norm property of amplitude encoding (Lipschitz constant 1) composed with observables whose operator norm is bounded by 1, but does not derive the composite Lipschitz constant or isolate the quantum circuit's contribution from the classical pre-encoder. In the revised manuscript we will add a dedicated subsection deriving an upper bound on the full-model Lipschitz constant and showing that the variational circuit with bounded observables further restricts sensitivity beyond the classical layer alone. revision: yes

  2. Referee: [Empirical Evaluation] Empirical results section: The claim of competitive performance and improved robustness/stability under noise and feature removal is presented without reported dataset sizes, number of independent runs, error bars, or statistical tests comparing the quantum model against baselines, undermining assessment of whether the data support the stated superiority in the SAFE dimensions.

    Authors: We concur that these details are necessary for rigorous evaluation. The current version omits explicit dataset sizes, the number of independent runs, error bars, and statistical comparisons. The revised manuscript will report the exact dataset sizes employed, results averaged over ten independent runs with standard-error bars, and the outcomes of paired statistical tests (t-tests or Wilcoxon signed-rank tests with p-values) between the quantum model and each classical baseline on the SAFE metrics. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected; derivation remains self-contained

full rationale

The abstract presents the core claim as following from the combination of normalized amplitude embeddings with bounded quantum observables, yielding a hypothesis class with controlled sensitivity. This is framed as an inductive property of the model architecture rather than a fitted parameter or self-referential definition. SAFE-AI metrics are explicitly tied to the external Cramer-von-Mises divergence, providing an independent statistical foundation. No equations or steps in the provided text reduce a prediction or uniqueness result back to the same fitted inputs or self-citations by construction. The learnable pre-encoding layer is introduced as a stabilizer but is not shown to make the stability claim tautological. Empirical comparisons to classical baselines further indicate external validation rather than internal re-labeling of results.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the unproven assertion that bounded quantum observables plus a learnable pre-layer produce controlled sensitivity, plus multiple fitted parameters whose values are not reported.

free parameters (1)
  • parameters of the learnable classical pre-encoding layer
    The layer is explicitly learnable, implying parameters optimized on data to stabilize the quantum embedding.
axioms (1)
  • domain assumption Normalized amplitude embeddings combined with bounded quantum observables produce a structured and smooth hypothesis class with controlled input sensitivity.
    This premise is invoked directly in the abstract to justify the model's reliability properties.

pith-pipeline@v0.9.0 · 5648 in / 1205 out tokens · 36862 ms · 2026-05-20T20:43:58.046554+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost Jcost uniqueness / washburn_uniqueness_aczel echoes
    ?
    echoes

    ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

    By combining normalized amplitude embeddings with bounded quantum observables, the resulting model induces a structured and smooth hypothesis class with controlled sensitivity to input variations.

  • IndisputableMonolith/Foundation/AlphaCoordinateFixation costAlphaLog_high_calibrated_iff / alpha_pin_under_high_calibration echoes
    ?
    echoes

    ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

    Since quantum circuits used in the hybrid architecture consist of unitary transformations, they preserve distances at the level of quantum states... the quantum feature map is 1-Lipschitz... the prediction function is constructed as a composition of normalized state preparation, norm-preserving unitary transformations, and bounded linear functionals.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

64 extracted references · 64 canonical work pages · 2 internal anchors

  1. [1]

    2010 , publisher=

    Quantum Computation and Quantum Information , author=. 2010 , publisher=

    work page 2010

  2. [2]

    Contemporary Physics , volume=

    An introduction to quantum machine learning , author=. Contemporary Physics , volume=. 2015 , publisher=

    work page 2015

  3. [3]

    Nature , volume=

    Quantum machine learning , author=. Nature , volume=. 2017 , publisher=

    work page 2017

  4. [4]

    Springer , year=

    Machine learning with quantum computers , author=. Springer , year=

    work page

  5. [5]

    Nature Reviews Physics , volume=

    Variational quantum algorithms , author=. Nature Reviews Physics , volume=. 2021 , publisher=

    work page 2021

  6. [6]

    A Simple Way to Initialize Recurrent Networks of Rectified Linear Units

    A simple way to initialize recurrent networks of rectified linear units , author=. arXiv preprint arXiv:1504.00941 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  7. [7]

    International Conference on Machine Learning , pages=

    Recurrent orthogonal networks and long-memory tasks , author=. International Conference on Machine Learning , pages=. 2016 , organization=

    work page 2016

  8. [8]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=

    Orthogonal deep neural networks , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , volume=. 2019 , publisher=

    work page 2019

  9. [9]

    International conference on machine learning , pages=

    Train faster, generalize better: Stability of stochastic gradient descent , author=. International conference on machine learning , pages=. 2016 , organization=

    work page 2016

  10. [10]

    Advances in neural information processing systems , volume=

    Spectrally-normalized margin bounds for neural networks , author=. Advances in neural information processing systems , volume=

    work page

  11. [11]

    International Conference on Machine Learning , pages=

    Cheap orthogonal constraints in neural networks: A simple parametrization of the orthogonal and unitary group , author=. International Conference on Machine Learning , pages=. 2019 , organization=

    work page 2019

  12. [12]

    Advances in Neural Information Processing Systems , volume=

    Can we gain more from orthogonality regularizations in training deep networks? , author=. Advances in Neural Information Processing Systems , volume=

    work page

  13. [13]

    Nature Computational Science , volume=

    The power of quantum neural networks , author=. Nature Computational Science , volume=. 2021 , publisher=

    work page 2021

  14. [14]

    Nature Communications , volume=

    Generalization in quantum machine learning from few training data , author=. Nature Communications , volume=. 2022 , publisher=

    work page 2022

  15. [15]

    Computational Intelligence and Neuroscience , volume=

    A new model for brain tumor detection using ensemble transfer learning and quantum variational classifier , author=. Computational Intelligence and Neuroscience , volume=. 2022 , publisher=

    work page 2022

  16. [16]

    A survey of

    Bauer, Stefan and Wiest, Roland and Nolte, Lutz. A survey of. Physics in Medicine and Biology , volume =

    work page

  17. [17]

    Expert Systems with Applications , volume =

    Babaei, Golnoosh and Giudici, Paolo and Raffinetti, Emanuela , title =. Expert Systems with Applications , volume =. 2025 , issn =

    work page 2025

  18. [18]

    Scientific Reports , volume=

    Enhancing multiclass brain tumor diagnosis using SVM and innovative feature extraction techniques , author=. Scientific Reports , volume=. 2024 , publisher=

    work page 2024

  19. [19]

    and Ameer, P

    Deepak, S. and Ameer, P. M. , title =. Journal of King Saud University--Computer and Information Sciences , volume =

    work page

  20. [20]

    and others , title =

    Esteva, Andre and Kuprel, Brett and Novoa, Roberto A. and others , title =. Nature , volume =

    work page

  21. [21]

    2024 , address =

    Felefly, Tony , title =. 2024 , address =

    work page 2024

  22. [22]

    Advances in Data Analysis and Classification , volume =

    Giudici, Paolo and Raffinetti, Emanuela , title =. Advances in Data Analysis and Classification , volume =. 2025 , doi =

    work page 2025

  23. [23]

    Brain tumor segmentation with deep neural networks , journal =

    Havaei, Mohammad and Davy, Axel and Warde. Brain tumor segmentation with deep neural networks , journal =. 2017 , doi =

    work page 2017

  24. [24]

    Deep learning trends for focal brain pathology segmentation in

    Havaei, Mohammad and Guizard, Nicolas and Larochelle, Hugo and Jodoin, Pierre. Deep learning trends for focal brain pathology segmentation in. Brainlesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries , publisher =. 2016 , doi =

    work page 2016

  25. [25]

    Proceedings of the

    He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian , title =. Proceedings of the

    work page

  26. [26]

    MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

    Howard, Andrew G. and Zhu, Menglong and Chen, Bo and others , title =. arXiv preprint arXiv:1704.04861 , year =

    work page internal anchor Pith review Pith/arXiv arXiv

  27. [27]

    Biomedical Signal Processing and Control , volume=

    HiFuse: Hierarchical multi-scale feature fusion network for medical image classification , author=. Biomedical Signal Processing and Control , volume=. 2024 , publisher=

    work page 2024

  28. [28]

    nnU-Net: A self-adapting framework for

    Isensee, Fabian and J. nnU-Net: A self-adapting framework for. Nature Methods , volume =

    work page

  29. [29]

    International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=

    nnu-net revisited: A call for rigorous validation in 3d medical image segmentation , author=. International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=. 2024 , organization=

    work page 2024

  30. [30]

    , title =

    Krizhevsky, Alex and Sutskever, Ilya and Hinton, Geoffrey E. , title =. Advances in Neural Information Processing Systems , address =

    work page

  31. [31]

    Menze, B. H. and Jakab, A. and Bauer, S. and others , title =. IEEE Transactions on Medical Imaging , volume =. 2015 , doi =

    work page 2015

  32. [32]

    , title =

    Pereira, Sérgio and Meier, Raphael and Alves, Victor and Silva, Carlos A. , title =

    work page

  33. [33]

    2024 , publisher =

    Rahman, Md Mizanur , title =. 2024 , publisher =. doi:10.17632/mk56jw9rns.1 , url =

    work page doi:10.17632/mk56jw9rns.1 2024

  34. [34]

    Journal of Computational Science , volume =

    Sajjad, Muhammad and Khan, Saeed and Muhammad, Kashif and others , title =. Journal of Computational Science , volume =

    work page

  35. [35]

    Physical Review Letters , volume =

    Schuld, Maria and Killoran, Nathan , title =. Physical Review Letters , volume =. 2019 , publisher =

    work page 2019

  36. [36]

    International Conference on Learning Representations , year =

    Simonyan, Karen and Zisserman, Andrew , title =. International Conference on Learning Representations , year =

    work page

  37. [37]

    Proceedings of the 3rd International Conference on Communication, Security, and Artificial Intelligence (ICCSAI) , pages =

    A Hybrid Classical-Quantum Model for Enhanced MRI-Based Brain Tumor Classification Using Transfer Learning and Quantum Neural Networks , author =. Proceedings of the 3rd International Conference on Communication, Security, and Artificial Intelligence (ICCSAI) , pages =. 2025 , organization =

    work page 2025

  38. [38]

    Proceedings of the 36th International Conference on Machine Learning , address =

    Tan, Mingxing and Le, Quoc , title =. Proceedings of the 36th International Conference on Machine Learning , address =

    work page

  39. [39]

    Biomedical Engineering Online , volume =

    Zhang, Yudong and Wu, Lenan and Zhang, Peng , title =. Biomedical Engineering Online , volume =

    work page

  40. [40]

    Expert Systems with Applications , volume=

    An intelligent driven deep residual learning framework for brain tumor classification using MRI images , author=. Expert Systems with Applications , volume=. 2023 , publisher=

    work page 2023

  41. [41]

    and Tootell, Geoffrey M

    Munnell, Alicia H. and Tootell, Geoffrey M. B. and Browne, Lynn E. and McEneaney, James , title =. American Economic Review , volume =

    work page

  42. [42]

    and Canner, Glenn B

    Berkovec, James A. and Canner, Glenn B. and Gabriel, Stuart A. and Hannan, Timothy H. , title =. Review of Economics and Statistics , volume =

    work page

  43. [43]

    2025 , month =

    Elzayn, Hadi and Freyaldenhoven, Simon and Kobler, Ryan and Shin, Minchul , title =. 2025 , month =

    work page 2025

  44. [44]

    Journal of King Saud University Computer and Information Sciences , volume =

    Fusion-aware quantum variational autoencoder for brain--heart signal modeling in mental health applications , author =. Journal of King Saud University Computer and Information Sciences , volume =

    work page

  45. [45]

    2025 , month =

    Jun, Alice , title =. 2025 , month =

    work page 2025

  46. [46]

    and Killoran, N

    Schuld, M. and Killoran, N. , title =. Physical Review Letters , volume =. 2019 , doi =

    work page 2019

  47. [47]

    Stein, S. A. and Lupo, C. and Farrelly, T. and O'Donnell, R. , title =. arXiv preprint arXiv:2210.00000 , year =

    work page arXiv

  48. [48]

    Financial Innovation , year =

    Hybrid Quantum Neural Networks with Amplitude Encoding: Advancing Recovery Rate Predictions , author =. Financial Innovation , year =. 2501.15828 , archivePrefix =

    work page arXiv

  49. [49]

    Scientific reports , volume=

    Deep learning-driven brain tumor classification and segmentation using non-contrast MRI , author=. Scientific reports , volume=. 2025 , publisher=

    work page 2025

  50. [50]

    Sensors , volume=

    A novel approach for brain tumor classification using an ensemble of deep and hand-crafted features , author=. Sensors , volume=. 2023 , publisher=

    work page 2023

  51. [51]

    Physical Review A , volume=

    Quantum circuit learning , author=. Physical Review A , volume=. 2018 , publisher=

    work page 2018

  52. [52]

    arXiv preprint arXiv:2003.01695 , year=

    Robust data encodings for quantum classifiers , author=. arXiv preprint arXiv:2003.01695 , year=

    work page arXiv 2003

  53. [53]

    ACM Transactions on Asian and Low-Resource Language Information Processing , volume=

    Supervised machine learning method for ontology-based financial decisions in the stock market , author=. ACM Transactions on Asian and Low-Resource Language Information Processing , volume=. 2023 , publisher=

    work page 2023

  54. [54]

    Selvaraju and Michael Cogswell and Abhishek Das and Ramakrishna Vedantam and Devi Parikh and Dhruv Batra , title =

    Ramprasaath R. Selvaraju and Michael Cogswell and Abhishek Das and Ramakrishna Vedantam and Devi Parikh and Dhruv Batra , title =. Proceedings of the IEEE International Conference on Computer Vision (ICCV) , year =

    work page

  55. [55]

    Empowering Brain Tumor Diagnosis through Explainable Deep Learning , volume =

    Li, Zhengkun and Dib, Omar , year =. Empowering Brain Tumor Diagnosis through Explainable Deep Learning , volume =. Machine Learning and Knowledge Extraction , doi =

    work page

  56. [56]

    Optics Communications , volume=

    Quantum convolutional neural network based on variational quantum circuits , author=. Optics Communications , volume=. 2024 , publisher=

    work page 2024

  57. [57]

    BMC Medical Imaging , volume=

    Medical image diagnosis based on adaptive Hybrid Quantum CNN , author=. BMC Medical Imaging , volume=. 2023 , publisher=

    work page 2023

  58. [58]

    Radio signal classification by adversarially robust quan- tum machine learning.arXiv preprint arXiv:2312.07821, 2023

    Radio signal classification by adversarially robust quantum machine learning , author=. arXiv preprint arXiv:2312.07821 , year=

    work page arXiv

  59. [59]

    Nature Computational Science , volume=

    Theory of overparametrization in quantum neural networks , author=. Nature Computational Science , volume=. 2023 , publisher=

    work page 2023

  60. [60]

    2023 2nd International Conference for Innovation in Technology (INOCON) , pages=

    Brain tumor classification using VGG 16, ResNet50, and inception V3 transfer learning models , author=. 2023 2nd International Conference for Innovation in Technology (INOCON) , pages=. 2023 , organization=

    work page 2023

  61. [61]

    International Challenge on Cross-Modality Domain Adaptation for Medical Image Segmentation , pages=

    Advanced tumor segmentation in medical imaging: An ensemble approach for brats 2023 adult glioma and pediatric tumor tasks , author=. International Challenge on Cross-Modality Domain Adaptation for Medical Image Segmentation , pages=. 2023 , publisher=

    work page 2023

  62. [62]

    Journal of Engineering , volume=

    Enhanced detection and segmentation of brain tumors using a dense BW-CNN approach , author=. Journal of Engineering , volume=. 2024 , publisher=

    work page 2024

  63. [63]

    Explaining and Harnessing Adversarial Examples , journal =

    Goodfellow, Ian and Shlens, Jonathon and Szegedy, Christian , year =. Explaining and Harnessing Adversarial Examples , journal =

    work page

  64. [64]

    Papernot, N., McDaniel, P

    Adversarial Robustness Toolbox v1.0.0 , author=. arXiv preprint arXiv:1807.01069 , year=

    work page arXiv