- Rating
- 3(+1)
- Confidence
- 3(−1)
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper introduces a data-to-readout interface to connect circuit structure with the representation available to projected quantum kernel learning. It derives bounds on the readout dimension, classical tractability, and tests selected predictions numerically. From a high-level perspective, understanding how much useful information remains accessible through a limited quantum readout is an interesting problem in quantum machine learning. However, I find that some parts of both the theoretical and numerical contributions are overstated, and the essential contributions may not really meet the bar of ICLR (details are provided in the weakness part) I therefore do not recommend acceptance in its current form.
Critical strengths
The paper addresses a worthwhile research question, with a clear problem formulation and a thorough discussion of related work. It presents theoretical results alongside numerical experiments that provide some support for the authors’ theoretical claims.
Critical weaknesses
-
The contribution beyond prior work is unclear. Several central observations have close precedents. Kübler et al. (2021) already discussed low-dimensional quantum kernels whose features can be classically hard to compute. Gan et al. (2023) studied feature representations and expressivity within a broad framework for quantum kernels. The connection between useful additional features and their correlation with the prediction residual also appears in Bang et al. (2026). Although these works are discussed, the substantial new insight beyond organizing existing ideas within the proposed interface model remains unclear, and it makes the current manuscript more like an incremental work.
-
Much of the analysis applies to a restricted projected kernel construction: data are encoded only in subsystem A, while B starts in a fixed state and receives input dependence through a fixed circuit. Moreover, the decomposition into fixed states \tau_s requires the controlled structure in Eq. (3), or compatible extensions. This decomposition does not generally apply when data are encoded directly into both subsystems or when the circuit lacks this structure. In particular, the broader scope of Theorem 1 should be distinguished from the restricted analysis based on \tau_s . The latter does not provide a general characterization of projected quantum kernels. Actually, Huang et al. 2021 did not give such structure assumptions, and this dramatically diminishes the significance and generality of the proposed results.
-
The theoretical analysis provides limited conceptual novelty, and the information-loss interpretation needs clarification. Theorem 1 follows from an operator Schmidt expansion and a dimension count, together with the established submultiplicativity of operator Schmidt rank. The efficient-evaluation part of Theorem 2 follows directly from K. These arguments are natural, but their implications seem largely expected. More importantly, the introduction asks, “How much input information can reach the readout,” suggesting an analysis of information lost through projection. However, M is the second-moment matrix of features already determined by the chosen encoder and projected kernel. It describes the resulting representation, but does not by itself measure what projection removes from the original representation.
-
The quantum-advantage results also appear to offer limited additional insight. The observation that a constant-dimensional readout can retain computational hardness is not particularly surprising: a standard BQP computation already makes its decision by measuring a single output qubit. Copying this output into a blank readout register gives a straightforward construction of this kind. Furthermore, the two learning separations in Corollary 4 inherit their quantum learners and classical lower bounds from Liu et al. (2021) and Molteni et al. (2026), as the authors explicitly acknowledge. The added contribution is to represent these existing separations through narrow readouts. Theorem 2 includes a separate construction based on the Goldreich–Levin hard-core predicate, but its conceptual advance over established hardness-based constructions needs a clearer explanation. I recognize the statistical role of D yet D alone cannot determine classical tractability or quantum advantage. It remains unclear what additional guidance its analysis provides for identifying projected kernels that offer a learning advantage.
-
The emphasis on 1000 qubits overstates the scale of the numerical evidence. In the 1000-qubit rank experiment, the largest tested interface has only five crossing gates, and both Q, G have rank 32. The tested circuits also admit efficient classical evaluation through low-width tensor-network contractions. This is a useful verification of the predicted rank bound, but it involves a highly structured circuit and a small readout feature space. It does not demonstrate learning with a computationally demanding 1000-qubit representation. The abstract should make these qualifications clearer, since emphasizing the total qubit count gives an exaggerated impression of the scale and difficulty of the experiment.
Questions for authors
Please refer to the weakness part.
16 comments
Response to Reviewer PPCN: Part2
4. Corollary 1 bounds the kernel-matrix rank from the circuit rather than from the number of measured observables as in [3], and Appendix C extends the residual criterion of Corollary 3 to refined features. [3] bounds this rank by the classical rank plus the number of measured observables. The circuits of Table R1 have no classical features, and then measuring all non-identity readout Pauli observables gives , or with the constant term. [3] also links useful extra features to the prediction residual, and the residual criterion of Corollary 3 likewise concerns appended features. In the coordinates of Eq. (4), however, adding CNOTs between the registers refines existing features rather than appending new ones. Appendix C therefore extends this residual criterion to the refinement under a compatible ridge penalty.
W2. Much of the analysis applies to a restricted projected kernel construction: data are encoded only in subsystem A, while B starts in a fixed state and receives input dependence through a fixed circuit. Moreover, the decomposition into fixed states requires the controlled structure in Eq. (3), or compatible extensions. This decomposition does not generally apply when data are encoded directly into both subsystems or when the circuit lacks this structure. In particular, the broader scope of Theorem 1 should be distinguished from the restricted analysis based on . The latter does not provide a general characterization of projected quantum kernels. Actually, Huang et al. 2021 did not give such structure assumptions, and this dramatically diminishes the significance and generality of the proposed results.
Response to W2. To show which results need the two restrictions named in the review, a blank readout and the controlled structure of Eq. (3), and which do not, we proved, as Proposition 3 in Appendix B.5 of the revised manuscript, a dimension bound that needs neither restriction, with as in Eq. (2) and the span of the data-encoded readout states . We also checked this bound numerically in Table 4 of the same appendix (Table R2 below), and listed the results that still need both restrictions in a new paragraph after Corollary 1 and in Table 2 of Appendix A.4. Overall, when the data enter each register separately before a fixed circuit, the dimension bound of Proposition 3 needs neither restriction, whereas the factorization into fixed states and the matrix , and the results built on this factorization, require both. Table R2 checks the first half of this statement. With one angle on each readout qubit, grows with the number of CNOTs from 81 to 192 and never exceeds the bound of Proposition 3. Below we derive this bound in part 1, list which results need the two restrictions in part 2, and contrast the bound with the geometric difference of [4] in part 3.
Table R2. Dimension with a data-encoded readout and CNOTs between two fixed random local layers ().
| Readout encoding | ||||
|---|---|---|---|---|
| angle on each qubit | 81 | 108 | 144 | 192 |
| angles on each qubit | 256 | 256 | 256 | 256 |
1. The bound of Proposition 3 does not require a blank readout. Suppose the readout starts in a data-encoded state , and the fixed circuit acts on . Expand the circuit into operator-Schmidt terms, with as in Eq. (2). Then is a sum of terms, each a fixed linear map of times an input-dependent coefficient. Hence , where is the span of the states . Table R2 evaluates this bound. With one angle on each readout qubit, equals at and grows to 192 at , within the bound. With angles, already reaches the maximum at . For controlled gates as in Eq. (3), decomposing over the control sectors replaces by and gives . The branch states then depend on and are no longer fixed (we write them here as ), but this bound still holds.
References
[3] Bang, Jeongho, et al. “Learning with active quantum subspaces: Scalable hybrid advantage without full quantum data-encoding.” arXiv preprint arXiv:2606.00932 (2026).
[4] Huang, Hsin-Yuan, et al. “Power of data in quantum machine learning.” Nature Communications 12.1 (2021): 2631.
Response to Reviewer PPCN: Part3
2. The analysis based on differs in scope from Theorem 1, and the revised manuscript states which results rely on . With a blank readout, Theorem 1 holds for any fixed circuit after the encoder. The fixed states and the fixed matrix , by contrast, additionally require Eq. (3) or the sector-preserving extensions of Appendix B. The results that rely on and are Corollary 1, the control-qubit sampling results of Sec. 4.3, the explicit form of Corollary 3, Theorem 2, and the controlled realization in Corollary 4. Theorem 1, Corollary 2, the bound of Appendix F, and the KRR risk bounds of Appendices E.2–E.3 do not rely on and . In particular, with a blank readout the bound of Appendix F holds for controlled gates with arbitrary local gates between them, so it does not need Eq. (3).
3. Unlike the geometric difference of [4], the bound of Proposition 3 is set by the circuit before any kernel entry is computed, and it needs neither restriction when the data are encoded in each register separately. The geometric difference of [4] compares the prediction-error bounds of a quantum and a classical kernel through a scalar computed from both evaluated kernel matrices, so the matrices must be computed first. On the other hand, when both restrictions hold, that is, with a blank readout and the controlled structure of Eq. (3), the blank readout gives , so the controlled-case bound of Proposition 3 gives the bound of Corollary 1. The results built on then also become available. In the revised manuscript, this comparison with [4] is in Appendix A.5.
References
[4] Huang, Hsin-Yuan, et al. “Power of data in quantum machine learning.” Nature Communications 12.1 (2021): 2631.
Response to Reviewer PPCN: Part1
Thank you for your thoughtful and constructive feedback! Your comments have helped us more clearly demonstrate our contribution beyond prior work and the scope of our results. We sincerely appreciate your valuable suggestions and address your questions point by point in the following eight comments.
W1. The contribution beyond prior work is unclear. Several central observations have close precedents. … Although these works are discussed, the substantial new insight beyond organizing existing ideas within the proposed interface model remains unclear, and it makes the current manuscript more like an incremental work.
Response to W1. To clarify what is new relative to prior work, we compared the dimension bound D≤2beff that Corollary 1 gives for controlled gates as in Eq. (3) with the bounds and feature counts of the three closest prior analyses [1,2,3] on the same circuits. In the revised manuscript, this comparison is in Appendix A.5 and in Table 8 of Appendix H.1.4 (Table R1 below), and the relation to [3] is discussed after Corollary 3. Overall, the comparison shows that the gates from the data register to the readout, with the local gates between them, set a bound on the dimension D of Eq. (1), and this bound can lie far below the readout-size bound 4nB. Specifically, on the circuits of Table R1 the prior bounds and counts stay near 4,096, while D equals 2b and grows only from 2 to 32. Below we explain in part 1 why D follows the gates between the registers rather than the readout size, and in parts 2 to 4 how Corollary 1 differs from each of [1], [2], and [3].
Table R1. Dimension D of Eq. (1) vs. feature counts of prior analyses (controlled circuits of Appendix G with six data and six readout qubits, m=512 multi-XOR training inputs).
1. D and the estimation risk follow the gates between the registers, not the readout size. Table R1 compares D with the bounds and counts of [1,2,3] on circuits that differ only in the number b of CNOTs. Those bounds and counts stay near 46=4,096, and even counting only the b readout qubits reached by the CNOTs, the readout-size bound is still 4b. In contrast, D=2b grows only from 2 to 32. Sec. 4.1 of our paper gives the same D=2b for N=20 to 1,000 qubits, so D does not grow with the readout. Accordingly, for targets in the feature span, the excess risk of kernel regression, averaged over label noise and training sets, follows D rather than the readout size and stays close to σ2D/m (last column of Table R1).
2. [1] uses the readout size 4nB as the dimension bound, whereas Corollary 1 tells from the circuit when D lies far below 4nB. [1] argues that low-dimensional kernels can have classically hard features, and Corollary 1 shows when a circuit gives such a low-dimensional kernel. The bound 2beff of Corollary 1 depends only on the controlled gates between the registers, not on the readout size.
3. Where [2] represents projected kernels in a Pauli or Mercer basis, Corollary 1 bounds the span dimension from the circuit before any feature is evaluated. In the Pauli basis, for example, 4,096 Pauli features have non-zero weight in Table R1 yet span only 2b dimensions. In the Mercer basis, Corollary 1 likewise limits the number of nonzero eigenvalues to 2beff.
References
[1] Kübler, Jonas, Simon Buchholz, and Bernhard Schölkopf. “The inductive bias of quantum kernels.” Advances in Neural Information Processing Systems 34 (2021): 12661–12673.
[2] Gan, Beng Yee, Daniel Leykam, and Supanut Thanasilp. “A unified framework for trace-induced quantum kernels.” arXiv preprint arXiv:2311.13552 (2023).
[3] Bang, Jeongho, et al. “Learning with active quantum subspaces: Scalable hybrid advantage without full quantum data-encoding.” arXiv preprint arXiv:2606.00932 (2026).