pith. sign in

arxiv: 2410.01252 · v2 · submitted 2024-10-02 · 🪐 quant-ph · cs.LG

Resource-efficient equivariant quantum convolutional neural networks

Koki Chinzei , Quoc Hoan Tran , Yasuhiro Endo , Hirotaka Oshima This is my paper

Pith reviewed 2026-05-23 20:28 UTC · model grok-4.3

classification 🪐 quant-ph cs.LG
keywords equivariant quantum neural networksquantum convolutional neural networksresource efficiencysymmetry preservationmeasurement parallelizationbarren plateausnoisy quantum classificationgroup theoretical encoding
0
0 comments X

The pith

Splitting the QCNN circuit at the pooling layer encodes general symmetries and parallelizes measurements to reduce resource needs by a factor of the qubit count.

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

The paper introduces an equivariant split-parallelizing QCNN that uses group theory to encode symmetries beyond translations. The circuit is split at the pooling layer in a way that preserves those symmetries, allowing parallel estimation of observables and gradients. This yields measurement efficiency gains scaling with qubit number, while the model maintains trainability, avoids barren plateaus, and shows strong generalization on noisy quantum classification tasks. A sympathetic reader would care because near-term devices face strict limits on total measurements, so any method that lowers that cost without sacrificing performance directly expands what can be run.

Core claim

Using a group-theoretical approach, the model encodes general symmetries by splitting the circuit at the pooling layer while preserving symmetry. This splitting structure effectively parallelizes QCNNs to improve measurement efficiency in estimating the expectation value of an observable and its gradient by order of the number of qubits. The model also exhibits high trainability and generalization performance, including the absence of barren plateaus. Numerical experiments demonstrate that the equivariant sp-QCNN can be trained and generalized with fewer measurement resources than a conventional equivariant QCNN in a noisy quantum data classification task.

What carries the argument

The symmetry-preserving split at the pooling layer, which parallelizes the circuit for joint estimation of observables and gradients.

If this is right

  • Measurement counts for expectation values and gradients scale down by a factor proportional to the number of qubits.
  • The training landscape remains free of barren plateaus.
  • Generalization holds in noisy quantum data classification tasks.
  • The construction applies to arbitrary symmetry groups through the group-theoretical encoding.
  • The split structure works for any pooling layer that respects the symmetry.

Where Pith is reading between the lines

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

  • The same splitting idea might be applied to other variational quantum circuits that contain pooling or measurement layers to gain similar efficiency without new hardware.
  • On physical devices the reduction in total shots could allow either larger batch sizes or more circuit repetitions within a fixed wall-clock budget.
  • If the symmetry preservation holds only approximately under hardware noise, the model could still outperform unsplit versions provided the parallelization benefit exceeds the fidelity loss.
  • The approach separates the cost of symmetry enforcement from the cost of depth, which may help when combining multiple symmetry constraints.

Load-bearing premise

Splitting the circuit at the pooling layer preserves the encoded symmetry for general groups while still allowing the parallelization benefit.

What would settle it

A direct comparison experiment in which the split model either requires at least as many total measurements as the unsplit equivariant QCNN to reach the same classification accuracy or produces outputs that violate the target symmetry on the test set.

Figures

Figures reproduced from arXiv: 2410.01252 by Hirotaka Oshima, Koki Chinzei, Quoc Hoan Tran, Yasuhiro Endo.

Figure 1
Figure 1. Figure 1: FIG. 1. Basic structures of (a) conventional QCNN [ [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Correct and wrong examples of circuit splitting for [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. An example of circuit splitting for [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. (a) 2 [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. (a)–(c) Changes in training loss, test loss, and test accuracy during training. The solid, dotted, and dashed lines [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. (a) Histograms for the estimated expectation value of [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. All subgroups of [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
read the original abstract

Equivariant quantum neural networks (QNNs) are promising variational models that exploit symmetries to improve machine learning capabilities. Despite theoretical developments in equivariant QNNs, their implementation on near-term quantum devices remains challenging due to limited computational resources. This study proposes a resource-efficient model of equivariant quantum convolutional neural networks (QCNNs) called equivariant split-parallelizing QCNN (sp-QCNN). Using a group-theoretical approach, we encode general symmetries into our model beyond the translational symmetry addressed by previous sp-QCNNs. We achieve this by splitting the circuit at the pooling layer while preserving symmetry. This splitting structure effectively parallelizes QCNNs to improve measurement efficiency in estimating the expectation value of an observable and its gradient by order of the number of qubits. Our model also exhibits high trainability and generalization performance, including the absence of barren plateaus. Numerical experiments demonstrate that the equivariant sp-QCNN can be trained and generalized with fewer measurement resources than a conventional equivariant QCNN in a noisy quantum data classification task. Our results contribute to the advancement of practical quantum machine learning algorithms.

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

3 major / 2 minor

Summary. The manuscript proposes the equivariant split-parallelizing QCNN (sp-QCNN), which encodes general group symmetries into QCNNs via a group-theoretical construction. The circuit is split at the pooling layer to enable parallel measurement of observables and gradients, yielding an efficiency gain of order the number of qubits while preserving equivariance. The model is claimed to exhibit high trainability (including absence of barren plateaus) and generalization; numerical experiments on a noisy quantum data classification task are presented to show that the sp-QCNN requires fewer measurement resources than a conventional equivariant QCNN.

Significance. If the symmetry-preservation claim holds for general groups, the work would supply a concrete route to reducing measurement overhead in equivariant QNNs on near-term hardware. The extension beyond translational symmetry is a clear advance over prior sp-QCNN constructions. The reported numerical results, if adequately documented, would provide useful evidence of practical trainability.

major comments (3)
  1. [§3.2] §3.2 (Splitting Construction): the statement that splitting at the pooling layer preserves the encoded symmetry for arbitrary groups is asserted without an explicit commutation relation or representation-theoretic verification that the independent sub-circuits and recombination operator jointly commute with the group action; the argument is immediate only for translations and does not generalize automatically to non-abelian cases.
  2. [§5] §5 (Numerical Experiments): the efficiency and generalization claims rest on experiments whose circuit sizes, noise models, number of shots, and statistical error bars are not reported, preventing quantitative assessment of whether the observed resource reduction is statistically significant or merely consistent with the central claim.
  3. [§4.1] §4.1 (Equivariance Proof): the parallelization benefit is quantified as “order of the number of qubits,” yet no explicit scaling relation or table compares the measurement cost of the split versus unsplit circuits for the same observable, leaving the precise resource saving unverified.
minor comments (2)
  1. [Figure 3] Figure 3: the caption does not state whether the plotted curves include error bars or how many independent runs underlie each point.
  2. [Eq. (12)] Notation: the symbol for the pooling operator is redefined between Eq. (12) and Eq. (15) without an explicit statement that the two usages coincide.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their thorough and constructive review. We address each major comment below and will revise the manuscript to incorporate the requested clarifications and additions.

read point-by-point responses

  1. Referee: [§3.2] §3.2 (Splitting Construction): the statement that splitting at the pooling layer preserves the encoded symmetry for arbitrary groups is asserted without an explicit commutation relation or representation-theoretic verification that the independent sub-circuits and recombination operator jointly commute with the group action; the argument is immediate only for translations and does not generalize automatically to non-abelian cases.

    Authors: We agree that an explicit verification is needed for general groups. In the revised manuscript we will add a representation-theoretic argument establishing the required commutation relations between the group action, the independent sub-circuits after splitting, and the recombination operator, thereby confirming equivariance preservation for both abelian and non-abelian groups. revision: yes

  2. Referee: [§5] §5 (Numerical Experiments): the efficiency and generalization claims rest on experiments whose circuit sizes, noise models, number of shots, and statistical error bars are not reported, preventing quantitative assessment of whether the observed resource reduction is statistically significant or merely consistent with the central claim.

    Authors: We accept that these experimental parameters must be documented. The revised Section 5 will report the circuit sizes, the precise noise models employed, the number of shots used for each expectation-value estimate, and statistical error bars on all plotted quantities to enable quantitative evaluation of the resource savings. revision: yes

  3. Referee: [§4.1] §4.1 (Equivariance Proof): the parallelization benefit is quantified as “order of the number of qubits,” yet no explicit scaling relation or table compares the measurement cost of the split versus unsplit circuits for the same observable, leaving the precise resource saving unverified.

    Authors: We will augment the manuscript with an explicit scaling derivation together with a table that directly compares the number of measurements required by the split and unsplit circuits for an identical observable, thereby making the claimed order-of-qubits improvement quantitatively verifiable. revision: yes

Circularity Check

0 steps flagged

No significant circularity; construction is independent of inputs.

full rationale

The paper proposes a new equivariant sp-QCNN architecture via group-theoretical encoding and circuit splitting at the pooling layer. Resource efficiency follows directly from the parallel measurement enabled by the split structure, with symmetry preservation built into the design rather than derived from fitted parameters or prior self-referential results. Numerical experiments provide independent validation. No quoted step reduces by construction to an input (e.g., no fitted quantity renamed as prediction, no load-bearing self-citation chain, no ansatz smuggled via citation). The derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available, so no concrete free parameters, axioms, or invented entities can be extracted from the text. The approach implicitly relies on standard assumptions of variational quantum circuits and representation theory of finite groups.

pith-pipeline@v0.9.0 · 5729 in / 1093 out tokens · 68292 ms · 2026-05-23T20:28:42.138468+00:00 · methodology

discussion (0)

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

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Iterative Quantum Feature Maps

    quant-ph 2025-06 unverdicted novelty 5.0

    IQFMs iteratively constructs deep quantum feature maps from shallow circuits via classical augmentation weights and contrastive layer-wise training, outperforming QCNNs on noisy quantum data and matching classical neu...

Reference graph

Works this paper leans on

71 extracted references · 71 canonical work pages · cited by 1 Pith paper · 3 internal anchors

  1. [1]

    the branches Q(ℓ) = {Q(ℓ) i }sℓ i=1

    Circuit splitting The first step is appropriately splitting the circuit such that ∀g ∈ G does not change the splitting structure, i.e. the branches Q(ℓ) = {Q(ℓ) i }sℓ i=1. With the aforementioned Eqs. (3) and (4), there are three requirements for the G- equivariant circuit splitting as follows:

    work page

  2. [2]

    G-invariance of circuit splitting: g(Q(ℓ)) = Q(ℓ) for ∀g ∈ G. (25)

    work page

  3. [3]

    Q(ℓ+1) i ⊆ Q(ℓ) j

    Branches do not merge: ∀i, ∃j s.t. Q(ℓ+1) i ⊆ Q(ℓ) j . (26)

    work page

  4. [4]

    (28) In particular, Eq

    Branches are a partition of qubits: Q(ℓ) i ∩ Q(ℓ) j = ∅ for i ̸= j, (27) [ i Q(ℓ) i = Qbit. (28) In particular, Eq. (25) means that any symmetry oper- ation can permute the branches but never modify the entire splitting structure, as illustrated in Fig. 2, which is essential for the equivariance. The subgroup method gives a systematic way of circuit split...

    work page

  5. [5]

    |G(q)|/|H(ℓ) λ (q)| = |G|/|H(ℓ) λ | for ∀q ∈ P (ℓ) λ , (34)

    work page

  6. [6]

    P (ℓ) is G-independent, (35)

    work page

  7. [7]

    The details are provided in Appendix A 1 c

    P (ℓ) is G-complete, (36) where we have defined P (ℓ) = F λ P (ℓ) λ . The details are provided in Appendix A 1 c. To summarize, satisfy- ing all requirements of Eqs. (25)–(28) necessitates care- ful selection of H(ℓ) and P(ℓ) to fulfill the conditions of 9 First Layer Second Layer Third Layer FIG. 3. An example of circuit splitting for D4 = {e, c4, (c4)2,...

    work page

  8. [8]

    (7) is satisfied

    Unitary operators After determining the circuit splitting at the ℓth layer, we need to design the unitary operator acting on each branch such that Eq. (7) is satisfied. However, the con- ventional methods, such as the twirling and nullspace methods, cannot be straightforwardly applied to design the unitary of the equivariant sp-QCNN because its split- tin...

    work page

  9. [9]

    Before moving on to the details, we summarize the requirements for the G-equivariant circuit splitting Q(ℓ) = {Q(ℓ) i }i as follows:

    Sufficient conditions for circuit splitting This section provides sufficient conditions for appro- priate circuit splitting, which forms the basis for the subgroup method. Before moving on to the details, we summarize the requirements for the G-equivariant circuit splitting Q(ℓ) = {Q(ℓ) i }i as follows:

    work page

  10. [10]

    G-invariance: g(Q(ℓ)) = Q(ℓ) for ∀g ∈ G. (A1)

    work page

  11. [11]

    Q(ℓ+1) i ⊆ Q(ℓ) j

    Branches do not merge: ∀i, ∃j s.t. Q(ℓ+1) i ⊆ Q(ℓ) j . (A2)

    work page

  12. [12]

    (A4) In what follows, we provide sufficient conditions for these three requirements one by one

    Branches are a partition of qubits Qbit = [n]: Q(ℓ) i ∩ Q(ℓ) j = ∅ for i ̸= j, (A3) [ i Q(ℓ) i = Qbit. (A4) In what follows, we provide sufficient conditions for these three requirements one by one. For convenience, we recall the following terms regard- ing the action of G on Qbit. Definition 1 (G-equivalence of qubits) . We say that qubits q1, q2 ∈ Qbit ...

    work page

  13. [13]

    (A1)–(A4)

    Systematic method of circuit splitting We are ready to present a systematic method of find- ing the circuit splitting that satisfies Eqs. (A1)–(A4). By Corollary 1, the subgroup method allows us to obtain the G-invariant branches Q(ℓ) = {Q(ℓ) λ,i} based on subgroups H(ℓ) = {H(ℓ) λ } and qubit subsets P(ℓ) = {P (ℓ) λ }. In addi- tion to the G-invariance, t...

    work page

  14. [14]

    H(ℓ+1) λ ≤ H(ℓ) λ′ and P (ℓ+1) λ ⊆ P (ℓ) λ′ , (A20)

    ∀λ, ∃λ′ s.t. H(ℓ+1) λ ≤ H(ℓ) λ′ and P (ℓ+1) λ ⊆ P (ℓ) λ′ , (A20)

    work page

  15. [15]

    G λ Pλ is G-independent and G-complete, (A22) where s(ℓ) λ = |G|/|H(ℓ) λ |

    |G(q)|/|H(ℓ) λ (q)| = s(ℓ) λ for ∀q ∈ P (ℓ) λ , (A21) 3. G λ Pλ is G-independent and G-complete, (A22) where s(ℓ) λ = |G|/|H(ℓ) λ |. In order to find H(ℓ) and P(ℓ) satisfying these condi- tions, we employ a brute-force method. This method be- gins with considering all subgroups of G and correspond- ing well-behaved qubits. For instance, Fig. 7 shows all s...

    work page

  16. [16]

    Merge two qubit subsets P1 and P2 if their accom- panying subgroups are the same: (H, P1), (H, P2) → (H, P1 ⊔ P2). (A23)

    work page

  17. [17]

    (A24) We can perform these operations repeatedly to construct the ℓth circuit splitting from the (ℓ + 1)th one

    Change a subgroup H1 to a larger one H2 (i.e., H1 ≤ H2): (H1, P) → (H2, P). (A24) We can perform these operations repeatedly to construct the ℓth circuit splitting from the (ℓ + 1)th one. This con- struction trivially satisfies Eq. (A20) and thus Eq. (A2). Furthermore, Eqs. (A21) and (A22) hold even after the above two operations. This is because F λ Pλ r...

    work page

  18. [18]

    The parameters are shared as δi 1,6 = δi 6,4 = δi 4,7 = δi 7,1, (B19) αi 1 = αi 4, (B20) αi 6 = αi

    (B16) The second convolutional layer V (2) = V (2) 2 V (2) 1 is given by V (2) 1 (θ) = d2Y i=1     Y ⟨j,k⟩∈P2 Rj,k(δi j,k)     Y j∈Q(2) 1 Rj(αi j)     , (B17) V (2) 2 (θ) = Uc4 V (2) 1 (θ)U † c4 , (B18) where we have defined P2 = {(1, 6), (6, 4), (4, 7), (7, 1)}. The parameters are shared as δi 1,6 = δi 6,4 = δi 4,7 = δi 7,1, (B19) αi 1 = ...

    work page

  19. [19]

    (B21) The third convolutional (or fully-connected) layer V (3) = V (3) 4 V (3) 3 V (3) 2 V (3) 1 is given by V (3) 1 (θ) = d3Y i=1 R1,4(δi 1,4)R1(αi 1)R4(αi 4), (B22) V (3) 2 (θ) = d3Y i=1 R6,7(δi 6,7)R6(αi 6)R7(αi 7), (B23) V (3) 3 (θ) = Uc4 V (3) 1 (θ)U † c4 , (B24) V (3) 4 (θ) = Uc4 V (3) 2 (θ)U † c4 , (B25) with parameter sharing αi 1 = αi 4, (B26) αi 6 = αi

    work page

  20. [20]

    P. W. Shor, Polynomial-Time Algorithms for Prime Fac- torization and Discrete Logarithms on a Quantum Com- puter, SIAM J. Comput. 26, 1484–1509 (1997)

    work page 1997

  21. [21]

    L. K. Grover, Quantum mechanics helps in searching for a needle in a haystack, Phys. Rev. Lett. 79, 325 (1997)

    work page 1997

  22. [22]

    How to factor 2048 bit rsa integers in 8 hours using 20 million noisy qubits,

    C. Gidney and M. Eker˚ a, How to factor 2048 bit RSA in- tegers in 8 hours using 20 million noisy qubits, Quantum 5, 433 (2021), 1905.09749v3

    work page arXiv 2048

  23. [23]

    Preskill, Quantum computing in the NISQ era and beyond, Quantum 2, 79 (2018)

    J. Preskill, Quantum computing in the NISQ era and beyond, Quantum 2, 79 (2018)

    work page 2018

  24. [24]

    Cerezo, A

    M. Cerezo, A. Arrasmith, R. Babbush, S. C. Benjamin, S. Endo, K. Fujii, J. R. McClean, K. Mitarai, X. Yuan, L. Cincio, and P. J. Coles, Variational quantum algo- rithms, Nat. Rev. Phys. 3, 625 (2021)

    work page 2021

  25. [25]

    Peruzzo, J

    A. Peruzzo, J. McClean, P. Shadbolt, M.-H. Yung, X.-Q. Zhou, P. J. Love, A. Aspuru-Guzik, and J. L. O’Brien, A variational eigenvalue solver on a photonic quantum processor, Nat. Commun. 5, 4213 (2014)

    work page 2014

  26. [26]

    Classification with Quantum Neural Networks on Near Term Processors

    E. Farhi and H. Neven, Classification with quantum neu- ral networks on near term processors, arXiv:1802.06002 [quant-ph] (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  27. [27]

    Mitarai, M

    K. Mitarai, M. Negoro, M. Kitagawa, and K. Fujii, Quan- tum circuit learning, Phys. Rev. A 98, 032309 (2018)

    work page 2018

  28. [28]

    Benedetti, E

    M. Benedetti, E. Lloyd, S. Sack, and M. Fiorentini, Pa- rameterized quantum circuits as machine learning mod- els, Quantum Sci. Technol. 4, 043001 (2019)

    work page 2019

  29. [29]

    Schuld, A

    M. Schuld, A. Bocharov, K. M. Svore, and N. Wiebe, Circuit-centric quantum classifiers, Phys. Rev. A 101, 032308 (2020)

    work page 2020

  30. [30]

    A Quantum Approximate Optimization Algorithm

    E. Farhi, J. Goldstone, and S. Gutmann, A quantum approximate optimization algorithm, arXiv:1411.4028 [quant-ph] (2014). 19

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  31. [31]

    J. R. McClean, S. Boixo, V. N. Smelyanskiy, R. Babbush, and H. Neven, Barren plateaus in quantum neural net- work training landscapes, Nat. Commun. 9, 4812 (2018)

    work page 2018

  32. [32]

    Cerezo, A

    M. Cerezo, A. Sone, T. Volkoff, L. Cincio, and P. J. Coles, Cost function dependent barren plateaus in shal- low parametrized quantum circuits, Nat. Commun. 12, 1791 (2021)

    work page 2021

  33. [33]

    Ortiz Marrero, M

    C. Ortiz Marrero, M. Kieferov´ a, and N. Wiebe, Entanglement-induced barren plateaus, PRX quantum2, 040316 (2021)

    work page 2021

  34. [34]

    Holmes, K

    Z. Holmes, K. Sharma, M. Cerezo, and P. J. Coles, Con- necting ansatz expressibility to gradient magnitudes and barren plateaus, PRX quantum 3, 010313 (2022)

    work page 2022

  35. [35]

    Larocca, S

    M. Larocca, S. Thanasilp, S. Wang, K. Sharma, J. Bia- monte, P. J. Coles, L. Cincio, J. R. McClean, Z. Holmes, and M. Cerezo, A Review of Barren Plateaus in Vari- ational Quantum Computing, arXiv:2405.00781 [quant- ph] (2024)

    work page arXiv 2024

  36. [36]

    M. M. Bronstein, J. Bruna, T. Cohen, and P. Veliˇ ckovi´ c, Geometric deep learning: Grids, groups, graphs, geodesics, and gauges, arXiv:2104.13478 [cs.LG] (2021)

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  37. [37]

    Verdon, T

    G. Verdon, T. McCourt, E. Luzhnica, V. Singh, S. Le- ichenauer, and J. Hidary, Quantum Graph Neural Net- works, arXiv:1909.12264 [quant-ph] (2019)

    work page arXiv 1909

  38. [38]

    Zheng, Z

    H. Zheng, Z. Li, J. Liu, S. Strelchuk, and R. Kon- dor, Speeding up learning quantum states through group equivariant convolutional quantum ans¨ atze, PRX quan- tum 4, 020327 (2023)

    work page 2023

  39. [39]

    Larocca, F

    M. Larocca, F. Sauvage, F. M. Sbahi, G. Verdon, P. J. Coles, and M. Cerezo, Group-invariant quantum machine learning, PRX quantum 3, 030341 (2022)

    work page 2022

  40. [40]

    J. J. Meyer, M. Mularski, E. Gil-Fuster, A. A. Mele, F. Arzani, A. Wilms, and J. Eisert, Exploiting symmetry in variational quantum machine learning, PRX quantum 4, 010328 (2023)

    work page 2023

  41. [41]

    Skolik, M

    A. Skolik, M. Cattelan, S. Yarkoni, T. B¨ ack, and V. Dunjko, Equivariant quantum circuits for learning on weighted graphs, arXiv:2205.06109 [quant-ph] (2022)

    work page arXiv 2022

  42. [42]

    Ragone, Q

    M. Ragone, P. Braccia, Q. T. Nguyen, L. Schatzki, P. J. Coles, F. Sauvage, M. Larocca, and M. Cerezo, Represen- tation theory for Geometric Quantum Machine Learning, arXiv:2210.07980 [quant-ph] (2022)

    work page arXiv 2022

  43. [43]

    Q. T. Nguyen, L. Schatzki, P. Braccia, M. Ragone, P. J. Coles, F. Sauvage, M. Larocca, and M. Cerezo, Theory for Equivariant Quantum Neural Networks, arXiv:2210.08566 [quant-ph] (2022)

    work page arXiv 2022

  44. [44]

    Sauvage, M

    F. Sauvage, M. Larocca, P. J. Coles, and M. Cerezo, Building spatial symmetries into parameterized quan- tum circuits for faster training, Quantum Sci. Technol. 9, 015029 (2024)

    work page 2024

  45. [45]

    Schatzki, M

    L. Schatzki, M. Larocca, Q. T. Nguyen, F. Sauvage, and M. Cerezo, Theoretical guarantees for permutation- equivariant quantum neural networks, arXiv:2210.09974 [quant-ph] (2022)

    work page arXiv 2022

  46. [46]

    Grant, M

    E. Grant, M. Benedetti, S. Cao, A. Hallam, J. Lockhart, V. Stojevic, A. G. Green, and S. Severini, Hierarchical quantum classifiers, Npj Quantum Inf. 4, 1 (2018)

    work page 2018

  47. [47]

    I. Cong, S. Choi, and M. D. Lukin, Quantum convolu- tional neural networks, Nat. Phys. 15, 1273 (2019)

    work page 2019

  48. [48]

    Pesah, M

    A. Pesah, M. Cerezo, S. Wang, T. Volkoff, A. T. Sorn- borger, and P. J. Coles, Absence of barren plateaus in quantum convolutional neural networks, Phys. Rev. X. 11, 041011 (2021)

    work page 2021

  49. [49]

    Schuld and N

    M. Schuld and N. Killoran, Is quantum advantage the right goal for quantum machine learning?, PRX quantum 3, 030101 (2022)

    work page 2022

  50. [50]

    Liu, Z.-Y

    H.-Y. Liu, Z.-Y. Chen, T.-P. Sun, C. Xue, Y.-C. Wu, and G.-P. Guo, Can variational quantum algorithms demonstrate quantum advantages? Time really matters, arXiv:2307.04089 [quant-ph] (2023)

    work page arXiv 2023

  51. [51]

    Chinzei, S

    K. Chinzei, S. Yamano, Q. H. Tran, Y. Endo, and H. Oshima, Trade-off between gradient measurement effi- ciency and expressivity in deep quantum neural networks, arXiv:2406.18316 [quant-ph] (2024)

    work page arXiv 2024

  52. [52]

    Chinzei, Q

    K. Chinzei, Q. H. Tran, K. Maruyama, H. Oshima, and S. Sato, Splitting and parallelizing of quantum convolu- tional neural networks for learning translationally sym- metric data, Phys. Rev. Res. 6, 023042 (2024)

    work page 2024

  53. [53]

    LeCun, Y

    Y. LeCun, Y. Bengio, and G. Hinton, Deep learning, Na- ture 521, 436 (2015)

    work page 2015

  54. [54]

    Krizhevsky, I

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, ImageNet classification with deep convolutional neural networks, Commun. ACM 60, 84 (2012)

    work page 2012

  55. [55]

    LeCun, Y

    Y. LeCun, Y. Bengio, and Others, Convolutional net- works for images, speech, and time series, The handbook of brain theory and neural networks 3361, 1995 (1995)

    work page 1995

  56. [56]

    Herrmann, S

    J. Herrmann, S. M. Llima, A. Remm, P. Zapletal, N. A. McMahon, C. Scarato, F. Swiadek, C. K. Andersen, C. Hellings, S. Krinner, N. Lacroix, S. Lazar, M. Ker- schbaum, D. C. Zanuz, G. J. Norris, M. J. Hartmann, A. Wallraff, and C. Eichler, Realizing quantum convo- lutional neural networks on a superconducting quantum processor to recognize quantum phases, ...

    work page 2022

  57. [57]

    Y.-J. Liu, A. Smith, M. Knap, and F. Pollmann, Model-independent learning of quantum phases of matter with quantum convolutional neural networks, arXiv:2211.11786 [quant-ph] (2022)

    work page arXiv 2022

  58. [58]

    Monaco, O

    S. Monaco, O. Kiss, A. Mandarino, S. Vallecorsa, and M. Grossi, Quantum phase detection generalization from marginal quantum neural network models, Phys. Rev. B. 107, L081105 (2023)

    work page 2023

  59. [59]

    S. Y.-C. Chen, T.-C. Wei, C. Zhang, H. Yu, and S. Yoo, Quantum convolutional neural networks for high energy physics data analysis, Phys. Rev. Res. 4, 013231 (2022)

    work page 2022

  60. [60]

    Nagano, A

    L. Nagano, A. Miessen, T. Onodera, I. Tavernelli, F. Tacchino, and K. Terashi, Quantum data learning for quantum simulations in high-energy physics, Phys. Rev. Res. 5, 043250 (2023)

    work page 2023

  61. [61]

    Schuld, V

    M. Schuld, V. Bergholm, C. Gogolin, J. Izaac, and N. Kil- loran, Evaluating analytic gradients on quantum hard- ware, Phys. Rev. A 99, 032331 (2019)

    work page 2019

  62. [62]

    Larocca, P

    M. Larocca, P. Czarnik, K. Sharma, G. Muraleedharan, P. J. Coles, and M. Cerezo, Diagnosing barren plateaus with tools from quantum optimal control, Quantum 6, 824 (2022), 2105.14377v3

    work page arXiv 2022

  63. [63]

    Ragone, B

    M. Ragone, B. N. Bakalov, F. Sauvage, A. F. Kemper, C. O. Marrero, M. Larocca, and M. Cerezo, A unified theory of barren plateaus for deep parametrized quantum circuits, arXiv:2309.09342 [quant-ph] (2023)

    work page arXiv 2023

  64. [64]

    Fontana, D

    E. Fontana, D. Herman, S. Chakrabarti, N. Kumar, R. Yalovetzky, J. Heredge, S. H. Sureshbabu, and M. Pis- toia, The adjoint is all you need: Characterizing Barren Plateaus in quantum ans¨ atze, arXiv:2309.07902 [quant- ph] (2023)

    work page arXiv 2023

  65. [65]

    J. J. Rotman, An introduction to the theory of groups , 4th ed., Graduate texts in mathematics (Springer, New 20 York, NY, 2012)

    work page 2012

  66. [66]

    D. P. Kingma and J. Ba, Adam: A method for stochastic optimization, in 3rd International Conference on Learn- ing Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, edited by Y. Bengio and Y. LeCun (2015)

    work page 2015

  67. [67]

    Robbins and S

    H. Robbins and S. Monro, A Stochastic Approximation Method, Ann. Math. Stat. 22, 400 (1951)

    work page 1951

  68. [68]

    Cerezo, M

    M. Cerezo, M. Larocca, D. Garc´ ıa-Mart´ ın, N. L. Diaz, P. Braccia, E. Fontana, M. S. Rudolph, P. Bermejo, A. Ijaz, S. Thanasilp, E. R. Anschuetz, and Z. Holmes, Does provable absence of barren plateaus imply clas- sical simulability? Or, why we need to rethink varia- tional quantum computing, arXiv:2312.09121 [quant-ph] (2023)

    work page arXiv 2023

  69. [69]

    Vidal, Entanglement renormalization, Phys

    G. Vidal, Entanglement renormalization, Phys. Rev. Lett. 99, 220405 (2007)

    work page 2007

  70. [70]

    Evenbly and G

    G. Evenbly and G. Vidal, Class of highly entangled many- body states that can be efficiently simulated, Phys. Rev. Lett. 112, 240502 (2014)

    work page 2014

  71. [71]

    Suzuki, Y

    Y. Suzuki, Y. Kawase, Y. Masumura, Y. Hiraga, M. Nakadai, J. Chen, K. M. Nakanishi, K. Mitarai, R. Imai, S. Tamiya, T. Yamamoto, T. Yan, T. Kawakubo, Y. O. Nakagawa, Y. Ibe, Y. Zhang, H. Yamashita, H. Yoshimura, A. Hayashi, and K. Fujii, Qulacs: a fast and versatile quantum circuit simulator for research pur- pose, Quantum 5, 559 (2021), 2011.13524v4

    work page arXiv 2021