Pith. sign in

REVIEW 3 major objections 5 minor 30 references

Component Based Quantum Machine Learning Explainability

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper answers yes to whether QML can be made explainable by explaining its core components, using fidelity-based SHAP and ALE to reveal feature and parameter importance that monolithic analysis hides.

desk verdict A plausible component-level decomposition for QML explainability, but the central claim rests on an unvalidated fidelity proxy and needs major revisions. read the letter →

arxiv 2506.12378 v1 pith:OJTKUSVG submitted 2025-06-14 quant-ph cs.AIcs.ETcs.LG

Barra White , Krishnendu Guha This is my paper
classification quant-phcs.AIcs.ETcs.LG PACS 03.67.Lx
keywords quantummachinelearningexplainabilitySHAPALEcomponent-basedframeworkstatefidelityvariationalclassifiersupportvector
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

This paper sets out to answer a yes-or-no question: can a quantum machine learning model be made explainable by explaining its constituent components instead of the whole model? It answers yes, by replacing the model's prediction with a quantum state fidelity measure and feeding that into two classical explainability tools, SHAP and ALE, separately for the feature map, the variational ansatz, the quantum kernel, and the classical decision function. Applied to a variational quantum classifier and a quantum support vector classifier on the Pima Indians diabetes dataset, the method surfaces component-specific feature and parameter importance, such as Age and BMI mattering mainly at the decision stage while Pregnancies matters mainly at the initial encoding stage. The paper positions this as the first component-based explainability framework for quantum machine learning.

What carries the argument

The central mechanism is the 'pseudo model': a wrapper that takes a quantum state (or a state produced by perturbed ansatz parameters) and returns its fidelity $F = |\langle \psi | \phi \rangle|^2$ to a reference state, typically the average encoded state of a class. This turns quantum states into scalar outputs that classical model-agnostic explainers such as SHAP (Shapley additive explanations, a game-theoretic feature attribution method) and ALE (accumulated local effects, a perturbation-based feature effect plot) can consume, at which point those methods compute Shapley values or accumulated local effects over the original input features for the feature map and kernel, or over the ansatz parameters for the variational circuit. For the quantum kernel, the technique instead evaluates kernel values against a central sample per class, and for the decision function a surrogate model is trained on the SVC decision output with an $R^2$ of 0.98.

What would settle it

Take a trained classifier and add a synthetic feature that is random noise but happens to push encoded states toward the class-average reference; if fidelity-based SHAP reports that noise feature as important while the model's accuracy is unchanged when the feature is dropped, the fidelity-to-class-average assumption fails. In other words, compare component-level SHAP importance with leave-one-feature-out model accuracy: any feature that is important in the fidelity plots but has no effect on accuracy would falsify the link between fidelity and prediction.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that explainability can be inferred for a QML algorithm by decomposing it into components and applying explainability techniques to each component independently. The key adaptation is the use of 'pseudo models' that output quantum state fidelity to a class-average state, so that SHAP and ALE, which normally require a scalar prediction, can be applied to quantum states and to ansatz parameters. The empirical results show that feature importance differs across components in a way that is invisible to monolithic analysis: for example, Glucose and SkinThickness retain similar importance across stages, but Age and BMI gain importance only in the decision function, while Pregnancies matters in the initial encoding but not in the final decision. The paper concludes that the answer to the question of whether component-based explainability works is yes, and claims this is the first such component-based technique.

Load-bearing premise

The entire pipeline assumes that the fidelity of a quantum state to a class-average reference state accurately represents how that component contributes to the model's prediction; if fidelity measures similarity to a constructed average rather than decision-relevant signal, the resulting plots are not explanations of the model.

Editorial extensions

If this is right

  • Feature engineering can now be targeted: features that matter only at the decision stage might be kept even if they look weak in the encoding, and features that matter only at encoding might be worth transforming or scaling.
  • Model tuning can become layer-specific: parameter-level SHAP values identify which ansatz layers or repetitions drive the output, enabling pruning or reinitialization of the least important parameters.
  • The modular framework is intended to be model-agnostic, so the same pseudo-model approach can be dropped into other QML algorithms and other components, such as different optimizers or quantum-classical loop stages.
  • Component-level importance can serve as a diagnostic for where a model is failing: if the encoding stage already separates the classes, the decision stage is the bottleneck, and vice versa.

Reading between the lines

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

  • Because the fidelity reference is built from class averages, the technique may conflate 'close to a typical example' with 'increases the predicted probability'; replacing the reference with a probability-weighted mixture of class states could separate these two notions.
  • A natural extension is to apply the same component decomposition to quantum regression, using target-value-conditioned reference states and inspecting how feature effects on the output emerge across components.
  • The component-level plots could be used to compare two feature maps directly: if one feature map assigns high importance to the same features in the decision function as the classical model, it may be a better inductive bias for that task.
  • The fidelity-based SHAP values for a single sample are local explanations; aggregating them across many samples could produce a global component-importance map that would let practitioners see, before training, which features the encoding will amplify.
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 / 5 minor

Summary. The paper proposes a modular framework for explainable quantum machine learning: instead of explaining a trained VQC or QSVC as a whole, it applies SHAP and ALE to individual components (feature map, ansatz, quantum kernel, SVC decision function) via 'pseudo-models' whose output is state fidelity to class-average quantum states. The framework is demonstrated on the Pima Indians Diabetes dataset, with waterfall plots, ALE curves, kernel eigenvalue/PCA analysis, and a surrogate-based SHAP analysis of the SVC decision function. The authors conclude that component-based explainability reveals feature and parameter importance that monolithic explainability would hide, answering their stated research question in the affirmative.

Significance. If the central assumption is validated, the modular decomposition is a plausible and potentially useful direction for XQAI: component-level attribution could guide feature-map and ansatz tuning in ways that monolithic explanations cannot. The paper has concrete strengths: it uses a real-world dataset, adapts two model-agnostic explainability methods, performs a surrogate analysis of the QSVC decision function with a reported R² of 0.98, and includes spectral analysis of the quantum kernel. However, the load-bearing claim depends on an unvalidated pseudo-model, and the evaluation is almost entirely qualitative, with no model accuracy numbers, no comparison to classical baselines despite a promise to do so, and no error bars or repeated-run statistics. The contribution is therefore promising but currently not established at the level required for a journal publication.

major comments (3)
  1. [Section IV and Section V.A] The pseudo-model that replaces the model prediction with fidelity to a 'class average quantum state' is never formally defined. If the class-average state is a normalized sum of statevectors, destructive interference can make it unrepresentative or nearly orthogonal to most class members; if it is a density-matrix mixture, the pseudo-model output becomes an average squared overlap, which is not the QSVC decision function (a kernel-weighted sum over support vectors) nor the VQC output (a parameterized measurement expectation). All SHAP and ALE plots in Figures 6-12 therefore explain similarity to a constructed reference, not necessarily the model's decision. The authors must provide evidence that the fidelity score tracks the actual decision function, for example by reporting the correlation or calibration between pseudo-model outputs and VQC/QSVC outputs on validation samples, or by an ablation showing that known decision-relevant features are recovered.
  2. [Section III.E and Section V] The paper promises 'a comprehensive and in depth analysis into the values will occur against classical ML algorithms', but no classical ML baseline results are reported, and no accuracy or F1 scores are given for the VQC or QSVC models themselves. Without these numbers, the conclusion that component-based explainability shows 'benefits' and 'the need' for the approach is not quantitatively supported. Please add a results table with VQC/QSVC and classical model metrics across repeated runs or seeds, and include quantitative explanation-quality measures such as agreement with the actual decision function or stability of SHAP values across runs.
  3. [Section IV.B and Section V.B.1] The ansatz SHAP adaptation is described only as 'slightly perturbing the original ansatz weights', but the perturbation scale, the number of SHAP samples, the background dataset, and the exact baseline for comparison are not specified. Likewise, the 'central sample from each class' used for kernel SHAP is undefined, and the ALE binning parameters are not given. These details are needed for reproducibility and for interpreting the parameter-importance plots in Figures 11 and 12. Please specify all hyperparameters of the explainability adaptations, or release the code in a way that makes the exact procedures available.
minor comments (5)
  1. [Section II.C.1 and Section V.A] Figure references are inconsistent: the ZZ feature map is called 'Figure 1' in the text but the actual circuit diagram is Figure 2, and several references appear as 'Figure ??' (e.g., for the ansatz circuit and the bloch sphere). Please correct these cross-references.
  2. [Throughout] There are numerous typos and grammatical slips that should be fixed, including 'alforithms' (Section II), 'od' (Section II.C), 'helper' (Section V.C.1), 'paperion' (Section V.C.1), and 'affect' where 'effect' is meant in the ALE description (Section II.H.2).
  3. [Section III.E] The sentence 'lit review table' appears to reference a table that is not present in the manuscript; either add the table or remove the dangling reference.
  4. [References] References [2] and [7] are duplicate entries for the same Gil-Fuster et al. paper, and some citations (e.g., [12] for the VQC diagram) do not appear to match the cited content. The bibliography should be cleaned and checked against the text.
  5. [Section VI] The claim that these are 'the first component based explainability techniques' is not supported by the related-work survey, which does not systematically demonstrate absence of prior component-level work. Please temper the claim or provide a more thorough literature search.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the component explanations are self-contained descriptions of an explicitly defined fidelity-based pseudo-model; the main weakness is construct validity, not circular reasoning.

full rationale

The paper does not derive a predictive result from a fitted input. Its adaptation replaces the model output with a stated target—state fidelity to class-average or baseline quantum states—and then runs off-the-shelf SHAP/ALE attributions on that explicitly defined pseudo-model. The resulting plots literally report sensitivity of the defined fidelity measure; no equation in the paper reduces a later claim to an earlier fitted constant. The central weakness is not circularity but construct validity: nothing in Sections IV and V shows that fidelity to a class-average state tracks the VQC/QSVC decision function, so the plots may explain the pseudo-model rather than the model. That is a correctness threat, not a self-referential derivation. The only self-citation ([26], Power and Guha) appears in the related-work section as a comparison point and is not load-bearing for the paper's framework, so it does not raise the circularity score.

Assumptions & free parameters 5 free parameters · 4 assumptions · 2 invented entities

The framework is built on two unvalidated constructs: the fidelity-based pseudo model and data-derived class-average reference states. It also imports the assumptions of SHAP and ALE unchanged and asserts compositional explainability without proof. No physical entities are invented, but the methodological constructs carry the main burden.

free parameters (5)
  • Z feature map repetition count = 2
    Selected because it yielded the highest VQC accuracy and F1 score in the authors' tests (Section II.C.1); those test metrics are not reported, and the explainability results depend on this choice.
  • Ansatz type and repetition count = EfficientSU2, repetitions unspecified
    The ansatz is chosen as a heuristic circuit and the text says only that the used circuit had "more repetitions" (Section V.B); without the exact structure, the parameter-importance SHAP values cannot be reproduced.
  • Optimizer choice = COBYLA
    COBYLA was chosen for low overhead and compatibility (Section II.E.2); changing the optimizer changes the trained ansatz parameters and therefore all downstream explanations.
  • Class-average quantum state baselines = not reported
    Fidelity targets for the pseudo model are averages of encoded class states computed from the training data (Section IV); explanations report similarity to these constructed references rather than to an externally grounded target.
  • Ansatz perturbation scale for SHAP = not reported
    The importance of ansatz parameters is measured by "slightly perturbing" the trained weights (Section IV.B); the perturbation size and sampling strategy are not specified.
assumptions (4)
  • domain assumption State fidelity to a class-average quantum state is a meaningful proxy for a component's contribution to model decisions.
    Invoked throughout Sections IV and V; no evidence links fidelity to class averages with actual prediction accuracy or decision boundaries.
  • domain assumption Explanations of individual QML components compose to explain the overall QML model.
    This is the central proposal in Sections I.A and VI, stated without proof or empirical check that component-level importances aggregate to global model behavior.
  • standard math Standard SHAP and ALE assumptions hold when the model output is a fidelity value rather than a prediction.
    The techniques are imported from the cited literature and applied to pseudo-models without re-deriving their assumptions for quantum state fidelity outputs.
  • domain assumption The Z feature map's class-average states are representative of the classes.
    The paper uses these averages as references without reporting separability or clustering metrics that would justify them as class prototypes (Section V.A).
invented entities (2)
  • Pseudo model (fidelity as prediction output)
    purpose: A callable that replaces a QML model's scalar prediction with the fidelity between a quantum state and a class-average reference state, so SHAP and ALE can be run on individual components.
    No external evidence connects fidelity-to-class-average to actual model decisions; it is a postulated proxy and the load-bearing target of every SHAP and ALE plot.
  • Class-average quantum state baseline
    purpose: Reference states representing each class, used to define the fidelity targets for the pseudo model.
    Computed from training data through the same feature map; explanations are relative to these constructed references, and their choice is not independently grounded.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Component Based Quantum Machine Learning Explainability." pith.science (2026). https://pith.science/paper/OJTKUSVG

@misc{pith2026250612378,
  author       = {Pith},
  title        = {Pith review of: Component Based Quantum Machine Learning Explainability},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OJTKUSVG}},
  note         = {Machine review of arXiv:2506.12378}
}
read the original abstract

Explainable ML algorithms are designed to provide transparency and insight into their decision-making process. Explaining how ML models come to their prediction is critical in fields such as healthcare and finance, as it provides insight into how models can help detect bias in predictions and help comply with GDPR compliance in these fields. QML leverages quantum phenomena such as entanglement and superposition, offering the potential for computational speedup and greater insights compared to classical ML. However, QML models also inherit the black-box nature of their classical counterparts, requiring the development of explainability techniques to be applied to these QML models to help understand why and how a particular output was generated. This paper will explore the idea of creating a modular, explainable QML framework that splits QML algorithms into their core components, such as feature maps, variational circuits (ansatz), optimizers, kernels, and quantum-classical loops. Each component will be analyzed using explainability techniques, such as ALE and SHAP, which have been adapted to analyse the different components of these QML algorithms. By combining insights from these parts, the paper aims to infer explainability to the overall QML model.

Figures

Figures reproduced from arXiv: 2506.12378 by the authors.

Figure 1
Figure 1. Circuit Diagram for a Z Feature Map with 2 repetitions. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Circuit Diagram for a ZZ Feature Map with linear entanglement and [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Circuit Diagram for a Real Amplitudes ansatz with linear entanglement [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗
Figures from the paper (12 more)
Figure 5
Figure 5. Figure 5: Diagram Showcasing how a VQC works. [12] [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: SHAP Waterfall showing what features push the quantum state towards [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: SHAP Waterfall showing what features push the quantum state towards [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: ALE plot for Glucose [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: ALE plot for SkinThickness. trained ansatz from the VQC. It changes up the classical SHAP by changing the players to be the trainable ansatz parameters. By comparing the output state of the trained ansatz to the perturbed version, we can get a SHAP value for each of t…
Figure 11
Figure 11. Figure 11: SHAP Waterfall plot showing parameter importance for the ansatz. [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]
Figure 12
Figure 12. Figure 12: SHAP Waterfall plot showing parameter layer importance for the [PITH_FULL_IMAGE:figures/full_fig_p008_12.png]
Figure 13
Figure 13. Figure 13: Eigenvalue Scree plot for first 20 components. [PITH_FULL_IMAGE:figures/full_fig_p009_13.png]
Figure 15
Figure 15. Figure 15: PCA projection for principal components 1, 2, and 3. [PITH_FULL_IMAGE:figures/full_fig_p009_15.png]
Figure 16
Figure 16. Figure 16: PCA projection for principal components 1, 2, and 4. [PITH_FULL_IMAGE:figures/full_fig_p009_16.png]
Figure 18
Figure 18. Figure 18: SHAP bar plot showing global feature importance for the decision [PITH_FULL_IMAGE:figures/full_fig_p010_18.png]
Figure 17
Figure 17. Figure 17: PCA projection for principal components 1, 2, and 3 showcasing [PITH_FULL_IMAGE:figures/full_fig_p010_17.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

30 extracted references · 26 canonical work pages

  1. [1]

    interpretability of black-box machine learning models,

    Please, explain. interpretability of black-box machine learning models,

    work page

  2. [3]

    Najafi, S

    K. Najafi, S. F. Yelin, and X. Gao. The development of quantum machine learning. Harvard Data Science Review , 2022. doi:10. 1162/99608f92.5a9fd72c

    work page 2022

  3. [4]

    Use of superposition and entanglement in quantum algorithms

    Sumit Asthana. Use of superposition and entanglement in quantum algorithms. URL: https://www.linkedin.com/pulse/ use-superposition-entanglement-quantum-algorithms-sumit-asthana

    work page

  4. [5]

    Hybrid quantum- classical models: Unlocking the power of quantum machine learning, 2024

    KqantumG Research Labs Pvt Ltd. Hybrid quantum- classical models: Unlocking the power of quantum machine learning, 2024. URL: https://www.linkedin.com/pulse/ hybrid-quantum-classical-models-unlocking-j0mxf

    work page 2024

  5. [6]

    Microsoft unveils majorana 1, the world’s first quantum processor powered by topological qubits, 2025

    Chetan Nayak. Microsoft unveils majorana 1, the world’s first quantum processor powered by topological qubits, 2025. URL: https://azure.microsoft.com/en-us/blog/quantum/2025/02/19/ microsoft-unveils-majorana-1-the-worlds-first-quantum-processor-powered-by-topological-qubits/ ?msockid=0e634514af3c6aef0cd050d3aed66b64

    work page 2025

  6. [7]

    Gil-Fuster, J

    E. Gil-Fuster, J. R. Naujoks, Wojciech Samek, and Jens Eisert. Oppor- tunities and limitations of explaining quantum machine learning. arXiv preprint, 2024. URL: https://arxiv.org/html/2412.14753v1

    arXiv 2024

  7. [8]

    Survey of encoding techniques for quantum machine learning

    Siddhartha Sharma and Renugadevi N. Survey of encoding techniques for quantum machine learning. Cybernetics and Physics Vol. 13 , 2024. URL: https://inspirehep.net/files/7a1f4eae2f5fadb4c1ea45dd0b884584

    work page 2024

  8. [9]

    Z feature map documentation

    IBM. Z feature map documentation. URL: https://docs.quantum.ibm. com/api/qiskit/qiskit.circuit.library.z feature map

    work page

Show all 30 references
  1. [10]

    Pauli feature map documentation

    IBM. Pauli feature map documentation. URL: https://docs.quantum. ibm.com/api/qiskit/qiskit.circuit.library.PauliFeatureMap

  2. [11]

    Qi et al

    H. Qi et al. Variational quantum algorithms: fundamental concepts, applications and challenges. Quantum Information Processing , 2024. doi:10.1007/s11128-024-04438-2

    2024 doi

  3. [12]

    Large scale diverse combinatorial optimization: Espn fantasy football player trades

    Aaron Baughman, Daniel Bohm, Micah Forster, Eduardo Morales, Jeff Powell, Shaun McPartlin, Raja Hebbar, and Kavitha Yogaraj. Large scale diverse combinatorial optimization: Espn fantasy football player trades. 11 2021. doi:10.48550/arXiv.2111.02859

    2021 doi arXiv

  4. [13]

    Variational algorithm design

    IBM Quantum Learning. Variational algorithm design. URL: https://learning.quantum.ibm.com/course/variational-algorithm-design/ variational-algorithms

  5. [14]

    Efficientsu2, 2025

    IBM. Efficientsu2, 2025. URL: https://docs.quantum.ibm.com/api/qiskit/ qiskit.circuit.library.EfficientSU2

    2025

  6. [15]

    Realamplitudes, 2025

    IBM. Realamplitudes, 2025. URL: https://docs.quantum.ibm.com/api/ qiskit/qiskit.circuit.library.RealAmplitudes

    2025

  7. [16]

    S. Faldu. Quantum support vector machine 101, 2023. URL: https: //dzone.com/articles/quantum-support-vector-machine-101

    2023

  8. [17]

    Navneet and S

    S. Navneet and S. Raj Pokhrel. An independent implementation of quantum machine learning algorithms in qiskit for genomic data. arXiv preprint, 2024. URL: https://arxiv.org/pdf/2405.09781/

    2024 arXiv

  9. [18]

    Quantum kernel machine learning, 2025

    IBM. Quantum kernel machine learning, 2025. URL: https://qiskit-community.github.io/qiskit-machine-learning/tutorials/ 03 quantum kernel.html

    2025

  10. [19]

    Explainable ai: Benefits and challenges, 2024

    BotPenguin. Explainable ai: Benefits and challenges, 2024. URL: https: //botpenguin.com/glossary/explainable-ai

    2024

  11. [20]

    Lundberg and Su-In Lee

    Scott M. Lundberg and Su-In Lee. A unified approach to interpreting model predictions. Conference on Neural Information Processing Systems, 2017

    2017

  12. [21]

    Accumulated local effects, 2020

    Oracle. Accumulated local effects, 2020. URL: https: //accelerated-data-science.readthedocs.io/en/v2.6.3/user guide/model explainability/accumulated local effects.html

    2020

  13. [22]

    Apley and Jingyu Zhu

    Daniel W. Apley and Jingyu Zhu. Visualizing the effects of predictor variables in black box supervised learning models. Journal of the Royal Statistical Society Series B: Statistical Methodology , 82, 2020. doi: 10.1111/rssb.12377

    2020 doi

  14. [23]

    Machine learning tutorials - qiskit machine learning 0.8.2

    IBM. Machine learning tutorials - qiskit machine learning 0.8.2. URL: https://qiskit-community.github.io/qiskit-machine-learning/ tutorials/index.html

  15. [24]

    Baughman, K

    A. Baughman, K. Yogaraj, R. Hebbar, S. Ghosh, R. U. Haq, and Y . Chhabra. Study of feature importance for quantum machine learning models. arXiv preprint, 2022. URL: https://arxiv.org/pdf/2202.11204

    2022 arXiv

  16. [25]

    Steinmuller, T

    P. Steinmuller, T. Schulz, F. Graf, and D. Herr. explainable ai for quantum machine learning. arXiv preprint , 2022. URL: https://arxiv. org/pdf/2211.01441

    2022 arXiv

  17. [26]

    Power and K

    L. Power and K. Guha. Feature importance and explainability in quantum machine learning. arXiv preprint , 2024. URL: https://arxiv. org/abs/2405.08917v1

    2024 arXiv

  18. [27]

    R. Heese. Explaining quantum circuits with shapley values: Towards explainable quantum machine learning. Quantum Machine Intelligence,

  19. [28]

    Abdulsalam, S

    G. Abdulsalam, S. Meshoul, and H. Shaiba. Explainable heart disease prediction using ensemble-quantum machine learning approach. Intelli- gent Automation and Soft Computing , 2022. doi:10.32604/iasc. 2023.032262

    2022

  20. [29]

    Qubit fidelity, 2025

    QuEra. Qubit fidelity, 2025. URL: https://www.quera.com/glossary/ fidelity

    2025

  21. [2019]

    URL: https://www.appsilon.com/post/please-explain-black-box

  22. [2025]

    doi:10.1007/s42484-025-00254-8

    doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.