pith. sign in

arxiv: 2509.06873 · v3 · submitted 2025-09-08 · 🪐 quant-ph

Entanglement and Classical Simulability in Quantum Extreme Learning Machines

A. De Lorenzis , M. P. Casado , N. Lo Gullo , T. Lux , F. Plastina , A. Riera This is my paper

Pith reviewed 2026-05-18 17:50 UTC · model grok-4.3

classification 🪐 quant-ph
keywords quantum extreme learning machinesentanglementclassical simulabilityXX Hamiltonianquantum machine learningfeature embeddingimage classification
0
0 comments X p. Extension

The pith

Moderate entanglement from local XX dynamics boosts QELM classification accuracy while remaining classically simulable.

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

This paper investigates quantum extreme learning machines that use dimensionality reduction, quantum encoding, evolution under an XX Hamiltonian, and measurements to generate features for classical classification. It finds that accuracy improves sharply as moderate entanglement develops, leading to better embedding of classical data in Hilbert space and more separable clusters in the measurement probabilities. The saturated performance reaches levels comparable to those from complex Haar-random unitaries despite the simple integrable dynamics. The evolution times involved correspond to short-distance effects and show no scaling with system size in the studied regimes, keeping the approach compatible with efficient classical simulation. This demonstrates that limited quantum correlations can enhance learnability without implying a need for quantum computational advantage.

Core claim

The central claim is that the increase in QELM performance correlates with the onset of entanglement under XX evolution, which improves the embedding of classical data in Hilbert space and produces more separable clusters in measurement probability space. For image classification tasks, this occurs at evolution times consistent with short-range information exchange that does not scale with system size, allowing the model to rely on limited entanglement while remaining classically simulable. The performance matches that achievable with maximally complex dynamics.

What carries the argument

Evolution under the XX Hamiltonian, which generates moderate short-range entanglement that structures the quantum feature representation for improved classical learnability.

If this is right

  • Classification accuracy exhibits a sharp transition to high values upon the onset of entanglement.
  • Saturated accuracy matches Haar-random unitary performance despite integrability of the XX model.
  • More separable clusters form in measurement probability space due to the entanglement-enhanced embedding.
  • The relevant evolution time remains independent of system size within tested scales, preserving classical simulability.

Where Pith is reading between the lines

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

  • If this pattern holds, then entanglement level could be tuned as a design parameter to balance performance and simulation cost in quantum learning models.
  • Classical simulation techniques focused on short-time local dynamics might replicate QELM benefits at larger scales.
  • Other quantum machine learning architectures might similarly benefit from moderate entanglement for feature enhancement without full quantum advantage.

Load-bearing premise

The relevant evolution times remain independent of system size and the resulting limited entanglement continues to suffice for performance at scales beyond those numerically tested.

What would settle it

A demonstration that for larger qubit numbers the time to reach high accuracy scales with system size and generates entanglement that is no longer efficiently classically simulable would falsify the simulability claim.

Figures

Figures reproduced from arXiv: 2509.06873 by A. De Lorenzis, A. Riera, F. Plastina, M. P. Casado, N. Lo Gullo, T. Lux.

Figure 1
Figure 1. Figure 1: Schematic representation of the QELM architecture. The workflow consists of the following steps: dimensionality reduction using either PCA or an Autoencoder (AE); encoding of the classical data into an initial quantum state; time evolution through the quantum layer; measurement of the evolved quantum state; and final classification via a classical single-layer neural network. Image for illustrative purpose… view at source ↗
Figure 2
Figure 2. Figure 2: Sample images from three widely used benchmark datasets in this study. The top row displays handwritten digits from MNIST (digits 0–9), the middle row shows clothing items from Fashion-MNIST (10 clothing categories), and the bottom row presents natural object categories from CIFAR-10. Each column corresponds to one class, labeled above each image. 10 3 10 2 10 1 10 0 10 1 10 2 t 0.75 0.80 0.85 0.90 0.95 1.… view at source ↗
Figure 3
Figure 3. Figure 3: Training (left panel) and testing (right panel) accuracy as a function of evolution time using the MNIST dataset. The time evolution and the encoding have been performed with the Hamiltonian XX model and the dense angle, respectively. The horizontal dashed lines indicate the performance obtained using a random unitary matrix, specifically by performing 10 measurements and computing the mean and standard de… view at source ↗
Figure 4
Figure 4. Figure 4: Training (left panel) and testing (right panel) accuracy as a function of evolution time using the Fashion-MNIST dataset. The time evolution and the encoding have been performed with the Hamiltonian XX model and the dense angle, respectively. The horizontal dashed lines indicate the performance obtained using a random unitary matrix, specifically by performing 10 measurements and computing the mean and sta… view at source ↗
Figure 5
Figure 5. Figure 5: Training (left panel) and testing (right panel) accuracy as a function of evolution time using the CIFAR-10 dataset. The time evolution and the encoding have been performed with the Hamiltonian XX model and the dense angle, respectively. The horizontal dashed lines indicate the performance obtained using a random unitary matrix, specifically by performing 10 measurements and computing the mean and standard… view at source ↗
Figure 6
Figure 6. Figure 6: a) Von Neumann entropy as function of the evolution time of half-chain (solid lines) and of the single qubit (dash-dotted lines). The colors denote the different sizes N (N = 6, 8, 10) of the entire system. b) Normalized (with N/2) von Neumann entropy as function of the evolution time (divided by N/2) of half-chain. 6 The Probability Polytope A natural framework for analyzing the dynamics of the Quantum Ex… view at source ↗
Figure 7
Figure 7. Figure 7: Accuracy (in magenta color) and Inertia (in light blue color) as a function of evolution time using the Fashion-MNIST dataset and N = 10. 10 2 10 1 10 0 10 1 10 2 t 80.0 80.5 81.0 81.5 82.0 82.5 83.0 Accuracy 20000 40000 60000 80000 Inertia Accuracy and Inertia vs t (Fashion-MNIST, N=10q with S i x , S i y , S i z ) (a) 10 2 10 1 10 0 10 1 10 2 t 80.0 80.5 81.0 81.5 82.0 82.5 83.0 Accuracy 0.275 0.300 0.32… view at source ↗
Figure 8
Figure 8. Figure 8: a) Accuracy (in magenta color) and Inertia (in light blue color) as a function of evolution time using the Fashion-MNIST dataset and N = 10. b) Accuracy (in magenta color) and ARI (in light green color) as a function of evolution time using the Fashion-MNIST dataset, for N = 10, obtained from the local expectation values only. New implementation (for odd qubits) encoding time evolution |0→ ω1, ω2 G G |0→ ω… view at source ↗
Figure 9
Figure 9. Figure 9: A schematic representation of an algorithm with 5 qubits, in which a unitary random matrix is applied for pairs of qubits contiguous. 10 [PITH_FULL_IMAGE:figures/full_fig_p010_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Training (left panel) and testing (right panel) accuracy as a function of the number of layers using the MNIST dataset. 1 2 3 4 5 6 N° Layers 0.25 0.30 0.35 0.40 0.45 0.50 Accuracy Accuracy vs N° Layers (Train - CIFAR10) Random Matrix N=6 Random Matrix N=7 Random Matrix N=8 Random Matrix N=9 Random Matrix N=10 Random Matrix N=11 N=6 N=7 N=8 N=9 N=10 N=11 (a) 1 2 3 4 5 6 N° Layers 0.25 0.30 0.35 0.40 0.45 … view at source ↗
Figure 11
Figure 11. Figure 11: Training (left panel) and testing (right panel) accuracy as a function of the number of layers using the CIFAR-10 dataset. 11 [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Lieb-Robinson light cone with velocity vLR = 2J. Outside this region, commutators of local observables are exponentially suppressed. A.3 Information Propagation from a Localized Excitation Consider a localized excitation created at t=0: |ψ(0)⟩ = c † j |0⟩. (17) Its time evolution: |ψ(t)⟩ = c † j (t)|0⟩, c † j (t)|0⟩ = X k i k−jJk−j (2J t)c † k , (18) shows that the amplitude on site k is |Jk−j (2J t)|, wh… view at source ↗
read the original abstract

Quantum Machine Learning (QML) has emerged as a promising framework for exploring how quantum dynamics may enhance data processing tasks. Here we investigate Quantum Extreme Learning Machines (QELMs), a quantum analogue of classical Extreme Learning Machines in which training is restricted to the output layer. Our architecture combines dimensionality reduction (via PCA or Autoencoders), quantum state encoding, evolution under an XX Hamiltonian, and projective measurement to produce features for a classical single-layer classifier. By analyzing the classification accuracy as a function of evolution time, we observe a sharp transition between low- and high-accuracy regimes, followed by saturation. Remarkably, the saturated performance is comparable to that obtained using Haar-random unitaries that generate maximally complex dynamics, even though the XX model is integrable and local. Our results indicate that this increase in performance correlates with the onset of entanglement, which improves the embedding of classical data in Hilbert space and leads to more separable clusters in measurement probability space. Thus, moderate entanglement can contribute positively to the structure of the data representation, improving learnability without necessarily implying quantum computational advantage. For the image-classification tasks studied here, namely MNIST, Fashion-MNIST, and CIFAR-10, the relevant evolution time is consistent with information exchange over short distances and, within the explored system sizes, does not show evidence of scaling with the full system size. This suggests that QELM performance in this regime relies only on limited entanglement and remains compatible with efficient classical simulation. Our results clarify how local quantum dynamics and moderate quantum correlations are already sufficient to generate useful feature representations for learning.

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 examines Quantum Extreme Learning Machines (QELMs) that encode classical image data (MNIST, Fashion-MNIST, CIFAR-10) via PCA or autoencoders, evolve the state under a local XX Hamiltonian, and extract features from projective measurements for a classical linear classifier. Numerical results show a sharp rise in classification accuracy with evolution time that saturates at values comparable to those obtained with Haar-random unitaries; this improvement correlates with the growth of entanglement entropy. The authors conclude that moderate, short-range entanglement suffices for useful feature maps and that the relevant evolution times remain independent of system size within the explored regimes, preserving classical simulability.

Significance. If the central numerical correlation holds and the limited-entanglement regime persists at scale, the work provides concrete evidence that local integrable dynamics can generate classically useful representations without requiring maximal entanglement or quantum advantage. The direct comparison to Haar-random unitaries and the explicit link to entanglement measures are strengths; the absence of free parameters in the core dynamical model further strengthens the result.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (results on scaling): the statement that 'the relevant evolution time is consistent with information exchange over short distances and, within the explored system sizes, does not show evidence of scaling with the full system size' is load-bearing for the classical-simulability claim, yet no scaling collapse, finite-size extrapolation, or analytic light-cone argument is supplied to support independence of t_sat from N in the thermodynamic limit. If t_sat grows even logarithmically, entanglement volume could become extensive and undermine both the 'moderate entanglement' and 'efficient classical simulation' conclusions.
  2. [§3.2] §3.2 (entanglement and accuracy plots): the reported correlation between entanglement onset and accuracy saturation is supported by direct simulation, but the manuscript does not quantify how much of the performance gain is attributable to entanglement versus other dynamical features (e.g., spreading of local operators) that would also appear in a classical tensor-network simulation of the same short-time dynamics.
minor comments (2)
  1. [Abstract] Abstract: no error bars, statistical tests, or data-exclusion criteria are mentioned for the accuracy-versus-time curves.
  2. [§3] Figure captions and §3: clarify whether the reported entanglement measures are averaged over the training set or computed on a single representative state.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments. We address each major comment below, indicating the revisions we will make to improve the manuscript.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (results on scaling): the statement that 'the relevant evolution time is consistent with information exchange over short distances and, within the explored system sizes, does not show evidence of scaling with the full system size' is load-bearing for the classical-simulability claim, yet no scaling collapse, finite-size extrapolation, or analytic light-cone argument is supplied to support independence of t_sat from N in the thermodynamic limit. If t_sat grows even logarithmically, entanglement volume could become extensive and undermine both the 'moderate entanglement' and 'efficient classical simulation' conclusions.

    Authors: We agree that a more rigorous justification for the lack of N-dependence in t_sat would strengthen the classical-simulability claim. In the revised manuscript we will add a finite-size scaling analysis of t_sat(N) using the system sizes already simulated plus additional accessible sizes, together with an explicit light-cone argument based on the Lieb-Robinson bound for the XX Hamiltonian. This bound establishes a finite propagation velocity, implying that the time required for short-distance information exchange remains O(1) and independent of N. We will incorporate these elements into §4 and update the abstract accordingly. While a complete extrapolation to the thermodynamic limit lies beyond the present computational scope, the added analysis will better support our statements within the regimes explored. revision: yes

  2. Referee: [§3.2] §3.2 (entanglement and accuracy plots): the reported correlation between entanglement onset and accuracy saturation is supported by direct simulation, but the manuscript does not quantify how much of the performance gain is attributable to entanglement versus other dynamical features (e.g., spreading of local operators) that would also appear in a classical tensor-network simulation of the same short-time dynamics.

    Authors: We thank the referee for this observation. While the manuscript shows a clear temporal correlation between entanglement entropy growth and accuracy saturation, we acknowledge that the relative contribution of entanglement versus classical operator spreading has not been quantified. In the revision we will add a comparison of the XX-evolved features to those obtained from a classical tensor-network simulation of the identical short-time dynamics (e.g., via matrix-product states with controlled bond dimension). This will allow us to isolate the additional benefit arising from entanglement. The new discussion and any supporting plots will be placed in §3.2. revision: yes

Circularity Check

0 steps flagged

No circularity: claims rest on direct numerical simulation of XX dynamics and entanglement measures

full rationale

The paper's central results on accuracy saturation, correlation with entanglement onset, and limited scaling of relevant evolution time are obtained from explicit numerical simulations on MNIST, Fashion-MNIST, and CIFAR-10. The abstract states that performance increases correlate with entanglement and that, within explored sizes, the time 'does not show evidence of scaling with the full system size.' This is an empirical observation from the simulations rather than any equation that defines the output in terms of itself or renames a fitted parameter as a prediction. No self-citations, uniqueness theorems, or ansatzes are invoked in the provided text to justify the architecture or conclusions. The derivation chain therefore remains independent of the target claims and is self-contained against the reported benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The work rests on standard quantum mechanics for state evolution and measurement; no new free parameters, ad-hoc axioms, or invented entities are introduced beyond the choice of the XX Hamiltonian and the numerical protocol.

axioms (1)
  • standard math Standard postulates of quantum mechanics govern unitary evolution under the XX Hamiltonian and projective measurements.
    Invoked throughout the description of state preparation, time evolution, and feature extraction.

pith-pipeline@v0.9.0 · 5834 in / 1447 out tokens · 62157 ms · 2026-05-18T17:50:53.076162+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.

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.

Forward citations

Cited by 2 Pith papers

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

  1. Theory and interpretability of Quantum Extreme Learning Machines: a Pauli-transfer matrix approach

    quant-ph 2026-02 unverdicted novelty 7.0

    A Pauli-transfer-matrix analysis of QELMs reveals the full set of nonlinear Pauli features generated by encoding and transformed by quantum channels, producing an interpretable classical nonlinear vector autoregressio...

  2. Optimal quantum reservoir learning in proximity to universality

    quant-ph 2025-10 unverdicted novelty 5.0

    A tunable mixing parameter p in random quantum circuits controls the transition from classically simulable to expressive quantum reservoir dynamics via entanglement and nonstabilizer content.

Reference graph

Works this paper leans on

69 extracted references · 69 canonical work pages · cited by 2 Pith papers · 1 internal anchor

  1. [1]

    Simulating physics with com- puters,

    R. P. Feynman, “Simulating physics with com- puters,”International journal of theoretical physics, vol. 21, no. 6/7, pp. 467–488, 1982

    work page 1982

  2. [2]

    Quantum algorithms: an overview,

    A. Montanaro, “Quantum algorithms: an overview,”npj Quantum Information, vol. 2, Jan. 2016

    work page 2016

  3. [3]

    M. A. Nielsen and I. L. Chuang,Quantum Computation and Quantum Information: 10th Anniversary Edition. Cambridge University Press, 2010

    work page 2010

  4. [4]

    Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer,

    P. W. Shor, “Polynomial-time algorithms for prime factorization and discrete logarithms on a quantum computer,”SIAM Journal on Computing, vol. 26, no. 5, pp. 1484–1509, 1997

    work page 1997

  5. [5]

    Quan- tum supremacy using a programmable su- perconducting processor,

    F. Arute, K. Arya, R. Babbush, D. Bacon, J. Bardin, R. Barends, R. Biswas, S. Boixo, F. Brandao, D. Buell, B. Burkett, Y. Chen, J. Chen, B. Chiaro, R. Collins, W. Court- ney, A. Dunsworth, E. Farhi, B. Foxen, A. Fowler, C. M. Gidney, M. Giustina, 14 R. Graff, K. Guerin, S. Habegger, M. Har- rigan, M. Hartmann, A. Ho, M. R. Hoff- mann, T. Huang, T. Humble,...

    work page 2019

  6. [6]

    Quantum computational advantage using photons,

    H.-S. Zhonget al., “Quantum computational advantage using photons,”Science, vol. 370, no. 6523, pp. 1460–1463, 2020

    work page 2020

  7. [7]

    Quantum advantage in learning from experiments,

    H.-Y. Huanget al., “Quantum advantage in learning from experiments,”Science, vol. 376, no. 6598, p. abn7293, 2022

    work page 2022

  8. [8]

    Quantum Computing in the NISQ era and beyond,

    J. Preskill, “Quantum Computing in the NISQ era and beyond,”Quantum, vol. 2, p. 79, 2018

    work page 2018

  9. [9]

    Quantum ma- chine learning,

    J. Biamonte, P. Wittek, N. Pancotti, P. Reben- trost, N. Wiebe, and S. Lloyd, “Quantum ma- chine learning,”Nature, vol. 549, no. 7671, pp. 195–202, 2017

    work page 2017

  10. [10]

    Schuld and F

    M. Schuld and F. Petruccione,Machine Learn- ing with Quantum Computers. Springer, 01 2021

    work page 2021

  11. [11]

    Schuld and F

    M. Schuld and F. Petruccione,Supervised Learning with Quantum Computers. Quan- tum Science and Technology, Springer, 2018

    work page 2018

  12. [12]

    Quan- tum convolutional neural networks,

    I. Cong, S. Choi, and M. Lukin, “Quan- tum convolutional neural networks,”Nature Physics, vol. 15, pp. 1–6, 12 2019

    work page 2019

  13. [13]

    Quantum capsule networks,

    Z. Liu, P.-X. Shen, W. Li, L. M. Duan, and D.- L. Deng, “Quantum capsule networks,”Quan- tum Sci. Technol., vol. 8, no. 1, p. 015016, 2023

    work page 2023

  14. [14]

    The quest for a quantum neural network,

    M. Schuld, I. Sinayskiy, and F. Petruccione, “The quest for a quantum neural network,” Quantum Information Processing, vol. 13, 08 2014

    work page 2014

  15. [15]

    Supervised learning with quantum-enhanced feature spaces,

    V. Havlicek, A. D. C´ orcoles, K. Temme, A. W. Harrow, A. Kandala, J. M. Chow, and J. M. Gambetta, “Supervised learning with quantum-enhanced feature spaces,”Nature, vol. 567, pp. 209–212, 2019

    work page 2019

  16. [16]

    Quantum algorithms for supervised and unsu- pervised machine learning,

    S. Lloyd, M. Mohseni, and P. Rebentrost, “Quantum algorithms for supervised and unsu- pervised machine learning,” 7 2013

    work page 2013

  17. [17]

    Assessing quan- tum computing performance for energy opti- mization in a prosumer community,

    C. Mastroianni, F. Plastina, L. Scarcello, J. Settino, and A. Vinci, “Assessing quan- tum computing performance for energy opti- mization in a prosumer community,”IEEE Transactions on Smart Grid, vol. 15, no. 1, pp. 444–456, 2023

    work page 2023

  18. [18]

    Variational quantum algorithms for the allocation of resources in a cloud/edge ar- chitecture,

    C. Mastroianni, F. Plastina, J. Settino, and A. Vinci, “Variational quantum algorithms for the allocation of resources in a cloud/edge ar- chitecture,”IEEE Transactions on Quantum Engineering, vol. 5, pp. 1–18, 2024

    work page 2024

  19. [19]

    Vari- ational gibbs state preparation on noisy intermediate-scale quantum devices,

    M. Consiglio, J. Settino, A. Giordano, C. Mas- troianni, F. Plastina, S. Lorenzo, S. Manis- calco, J. Goold, and T. J. Apollaro, “Vari- ational gibbs state preparation on noisy intermediate-scale quantum devices,”Phys- ical Review A, vol. 110, no. 1, p. 012445, 2024

    work page 2024

  20. [20]

    State estima- tion with quantum extreme learning machines beyond the scrambling time,

    M. Vetrano, G. Lo Monaco, L. Innocenti, S. Lorenzo, and G. M. Palma, “State estima- tion with quantum extreme learning machines beyond the scrambling time,” 9 2024

    work page 2024

  21. [21]

    Opportunities in quantum reservoir computing and extreme learning machines,

    P. Mujal, R. Mart´ ınez-Pe˜ na, J. Nokkala, J. Garc´ ıa-Beni, G. Giorgi, M. Soriano, and R. Zambrini, “Opportunities in quantum reservoir computing and extreme learning machines,”Advanced Quantum Technologies, vol. 4, 06 2021

    work page 2021

  22. [22]

    Po- tential and limitations of quantum extreme learning machines,

    L. Innocenti, S. Lorenzo, I. Palmisano, A. Fer- raro, M. Paternostro, and G. M. Palma, “Po- tential and limitations of quantum extreme learning machines,”Communications Physics, vol. 6, p. 118, May 2023

    work page 2023

  23. [23]

    On fundamental aspects of quantum extreme learning machines,

    W. Xiong, G. Facelli, M. Sahebi, O. Agnel, T. Chotibut, S. Thanasilp, and Z. Holmes, “On fundamental aspects of quantum extreme learning machines,” 12 2023

    work page 2023

  24. [24]

    Quantum Extreme Reservoir Computation Utilizing Scale-Free Networks,

    A. Sakurai, M. P. Estarellas, W. J. Munro, and K. Nemoto, “Quantum Extreme Reservoir Computation Utilizing Scale-Free Networks,” Phys. Rev. Applied, vol. 17, no. 6, p. 064044, 2022

    work page 2022

  25. [25]

    Effective quantum feature maps in quantum extreme reservoir computation 15 from the XY model,

    A. Hayashi, A. Sakurai, W. J. Munro, and K. Nemoto, “Effective quantum feature maps in quantum extreme reservoir computation 15 from the XY model,”Phys. Rev. A, vol. 111, no. 2, p. 022431, 2025

    work page 2025

  26. [26]

    Harnessing quantum extreme learning machines for image classi- fication,

    A. De Lorenzis, M. P. Casado, M. P. Estarel- las, N. L. Gullo, T. Lux, F. Plastina, A. Ri- era, and J. Settino, “Harnessing quantum extreme learning machines for image classi- fication,”Phys. Rev. Applied, vol. 23, no. 4, p. 044024, 2025

    work page 2025

  27. [27]

    Ex- perimental property reconstruction in a pho- tonic quantum extreme learning machine,

    A. Suprano, D. Zia, L. Innocenti, S. Lorenzo, V. Cimini, T. Giordani, I. Palmisano, E. Polino, N. Spagnolo, F. Sciarrino, G. M. Palma, A. Ferraro, and M. Paternostro, “Ex- perimental property reconstruction in a pho- tonic quantum extreme learning machine,” Phys. Rev. Lett., vol. 132, p. 160802, Apr 2024

    work page 2024

  28. [28]

    Large-scale quantum reservoir learning with an analog quantum computer,

    M. Kornjaˇ caet al., “Large-scale quantum reservoir learning with an analog quantum computer,” 7 2024

    work page 2024

  29. [29]

    Temporal information processing on noisy quantum computers,

    J. Chen, H. I. Nurdin, and N. Yamamoto, “Temporal information processing on noisy quantum computers,”Phys. Rev. Appl., vol. 14, p. 024065, Aug 2020

    work page 2020

  30. [30]

    Harnessing Quan- tum Dynamics for Robust and Scalable Quan- tum Extreme Learning Machines,

    P. D. Solanki and A. Pham, “Harnessing Quan- tum Dynamics for Robust and Scalable Quan- tum Extreme Learning Machines,” 3 2025

    work page 2025

  31. [31]

    Extreme learning machine: a new learning scheme of feedforward neural networks,

    G.-B. Huang, Q.-Y. Zhu, and C.-K. Siew, “Extreme learning machine: a new learning scheme of feedforward neural networks,” in 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541), vol. 2, pp. 985–990 vol.2, 2004

    work page 2004

  32. [32]

    Trends in extreme learning machines: A re- view,

    G. Huang, G.-B. Huang, S. Song, and K. You, “Trends in extreme learning machines: A re- view,”Neural Networks, vol. 61, pp. 32–48, 2015

    work page 2015

  33. [33]

    Extreme learning machine: algorithm, theory and applications,

    S. Ding, H. Zhao, Y. Zhang, X. Xu, and R. Nie, “Extreme learning machine: algorithm, theory and applications,”Artif. Intell. Rev., vol. 44, p. 103–115, June 2015

    work page 2015

  34. [34]

    A review on extreme learning machine,

    J. Wang, S. Lu, S.-H. Wang, and Y.-D. Zhang, “A review on extreme learning machine,”Mul- timedia Tools and Applications, vol. 81, no. 29, pp. 41611–41660, 2022

    work page 2022

  35. [35]

    Extreme learning machines: a survey. int j mach learn cybern,

    G.-B. Huang, D. Wang, and Y. Lan, “Extreme learning machines: a survey. int j mach learn cybern,”International Journal of Machine Learning and Cybernetics, vol. 2, pp. 107–122, 06 2011

    work page 2011

  36. [36]

    Simple hamiltonian dynamics as a powerful resource for image classification,

    A. Sakurai, A. Hayashi, W. J. Munro, and K. Nemoto, “Simple hamiltonian dynamics as a powerful resource for image classification,” Phys. Rev. A, vol. 111, p. 052432, May 2025

    work page 2025

  37. [37]

    Impact of the form of weighted networks on the quantum extreme reservoir computation,

    A. Hayashi, A. Sakurai, S. Nishio, W. J. Munro, and K. Nemoto, “Impact of the form of weighted networks on the quantum extreme reservoir computation,”Phys. Rev. A, vol. 108, no. 4, p. 042609, 2023

    work page 2023

  38. [38]

    Harnessing disordered-ensemble quantum dynamics for machine learning,

    K. Fujii and K. Nakajima, “Harnessing disordered-ensemble quantum dynamics for machine learning,”Physical Review Applied, vol. 8, p. 024030, 8 2017

    work page 2017

  39. [39]

    Reservoir computing approach to quantum state measurement,

    G. Angelatos, S. A. Khan, and H. E. T¨ ureci, “Reservoir computing approach to quantum state measurement,”Phys. Rev. X, vol. 11, p. 041062, Dec 2021

    work page 2021

  40. [40]

    An- alytical evidence of nonlinearity in qubits and continuous-variable quantum reservoir computing,

    P. Mujal, J. Nokkala, R. Mart´ ınez-Pe˜ na, G. L. Giorgi, M. C. Soriano, and R. Zambrini, “An- alytical evidence of nonlinearity in qubits and continuous-variable quantum reservoir computing,”Journal of Physics: Complexity, vol. 2, p. 045008, nov 2021

    work page 2021

  41. [41]

    Configured quantum reservoir computing for multi-task machine learning,

    W. Xia, J. Zou, X. Qiu, F. Chen, B. Zhu, C. Li, D.-L. Deng, and X. Li, “Configured quantum reservoir computing for multi-task machine learning,”Science Bulletin, vol. 68, pp. 2321–2329, 10 2023

    work page 2023

  42. [42]

    Dissipation as a resource for quantum reservoir computing,

    A. Sannia, R. Mart´ ınez-Pe˜ na, M. C. Soriano, G. L. Giorgi, and R. Zambrini, “Dissipation as a resource for quantum reservoir computing,” Quantum, vol. 8, p. 1291, 3 2024

    work page 2024

  43. [43]

    Dynam- ical phase transitions in quantum reservoir computing,

    R. Mart´ ınez-Pe˜ na, G. L. Giorgi, J. Nokkala, M. C. Soriano, and R. Zambrini, “Dynam- ical phase transitions in quantum reservoir computing,”Physical Review Letters, vol. 127, p. 100502, 9 2021

    work page 2021

  44. [44]

    Exploring quantum mechanical advantage for reservoir computing,

    N. G¨ otting, F. Lohof, and C. Gies, “Exploring quantum mechanical advantage for reservoir computing,”Physical Review A, vol. 108, 2 2023

    work page 2023

  45. [45]

    Informa- tion processing capacity of spin-based quan- tum reservoir computing systems,

    R. Mart´ ınez-Pe˜ na, J. Nokkala, G. L. Giorgi, R. Zambrini, and M. C. Soriano, “Informa- tion processing capacity of spin-based quan- tum reservoir computing systems,”Cognitive Computation, vol. 15, pp. 1440–1451, 9 2023

    work page 2023

  46. [46]

    Quantum reservoir computing for speckle disorder potentials,

    P. Mujal, “Quantum reservoir computing for speckle disorder potentials,”Condensed Mat- ter, vol. 7, p. 17, 1 2022. 16

    work page 2022

  47. [47]

    Learning nonlinear input–output maps with dissipative quantum systems,

    J. Chen and H. I. Nurdin, “Learning nonlinear input–output maps with dissipative quantum systems,”Quantum Information Processing, vol. 18, no. 7, p. 198, 2019

    work page 2019

  48. [48]

    Boosting computational power through spatial multiplexing in quan- tum reservoir computing,

    K. Nakajima, K. Fujii, M. Negoro, K. Mitarai, and M. Kitagawa, “Boosting computational power through spatial multiplexing in quan- tum reservoir computing,”Physical Review Applied, vol. 11, no. 3, p. 034021, 2019

    work page 2019

  49. [49]

    Higher-order quantum reservoir computing,

    Q. H. Tran and K. Nakajima, “Higher-order quantum reservoir computing,”arXiv preprint arXiv:2006.08999, 2020

    work page arXiv 2006

  50. [50]

    Informa- tion processing capacity of spin-based quan- tum reservoir computing systems,

    R. Mart´ ınez-Pe˜ na, J. Nokkala, G. L. Giorgi, R. Zambrini, and M. C. Soriano, “Informa- tion processing capacity of spin-based quan- tum reservoir computing systems,”Cognitive Computation, pp. 1–12, 2020

    work page 2020

  51. [51]

    Memory-augmented hybrid quantum reservoir computing,

    J. Settinoet al., “Memory-augmented hybrid quantum reservoir computing,”Phys. Rev. Ap- plied, vol. 24, no. 2, p. 024019, 2025

    work page 2025

  52. [52]

    Quantum reservoir processing,

    S. Ghosh, A. Opala, M. Matuszewski, T. Pa- terek, and T. C. H. Liew, “Quantum reservoir processing,”npj Quantum Information, vol. 5, p. 35, 4 2019

    work page 2019

  53. [53]

    Quantum-classical hybrid information process- ing via a single quantum system,

    Q. H. Tran, S. Ghosh, and K. Nakajima, “Quantum-classical hybrid information process- ing via a single quantum system,”Physical Review Research, vol. 5, p. 43127, 2023

    work page 2023

  54. [54]

    Benchmarking the role of par- ticle statistics in quantum reservoir comput- ing,

    G. Llodr` a, C. Charalambous, G. L. Giorgi, and R. Zambrini, “Benchmarking the role of par- ticle statistics in quantum reservoir comput- ing,”Advanced Quantum Technologies, vol. 6, p. 2200100, 1 2023

    work page 2023

  55. [55]

    Large-scale quantum reservoir learning with an analog quantum computer,

    M. Kornjaˇ ca, H.-Y. Hu, C. Zhao, J. Wurtz, P. Weinberg, M. Hamdan, A. Zhdanov, S. H. Cantu, H. Zhou, R. A. Bravo, K. Bagnall, J. I. Basham, J. Campo, A. Choukri, R. Deangelo, P. Frederick, D. Haines, J. Hammett, N. Hsu, M.-G. Hu, F. Huber, P. N. Jepsen, N. Jia, T. Karolyshyn, M. Kwon, J. Long, J. Lopatin, A. Lukin, T. M. Macr‘ı, O. M. Markovi´c, L. A. Ma...

    work page 2024

  56. [56]

    Quantum reservoir computing using arrays of rydberg atoms,

    R. A. Bravo, K. Najafi, X. Gao, and S. F. Yelin, “Quantum reservoir computing using arrays of rydberg atoms,”PRX Quantum, vol. 3, p. 030325, 8 2022

    work page 2022

  57. [57]

    Quantum computing with neu- tral atoms,

    L. Henriet, L. Beguin, A. Signoles, T. La- haye, A. Browaeys, G.-O. Reymond, and C. Jurczak, “Quantum computing with neu- tral atoms,”Quantum, vol. 4, p. 327, 9 2020

    work page 2020

  58. [58]

    Predicting fermionic densities using a projected-quantum-kernel method,

    F. Perciavalle, F. Plastina, M. Pisarra, and N. L. Gullo, “Predicting fermionic densities using a projected-quantum-kernel method,” Phys. Rev. A, vol. 112, no. 2, p. 022620, 2025

    work page 2025

  59. [59]

    Quantum superpositions of current states in rydberg-atom networks,

    F. Perciavalle, D. Rossini, J. Polo, O. Morsch, and L. Amico, “Quantum superpositions of current states in rydberg-atom networks,” Phys. Rev. Res., vol. 6, p. 043025, Oct 2024

    work page 2024

  60. [60]

    Autoassociative neural networks,

    M. Kramer, “Autoassociative neural networks,” Computers & Chemical Engineering, vol. 16, no. 4, pp. 313–328, 1992. Neutral network applications in chemical engineering

    work page 1992

  61. [61]

    Autoencoders, minimum description length and helmholtz free energy,

    G. E. Hinton and R. S. Zemel, “Autoencoders, minimum description length and helmholtz free energy,” inNeural Information Processing Systems, 1993

    work page 1993

  62. [62]

    Learning representations by back- propagating errors,

    D. E. Rumelhart, G. E. Hinton, and R. J. Williams, “Learning representations by back- propagating errors,”Nature, vol. 323, pp. 533– 536, 1986

    work page 1986

  63. [63]

    Re- ducing the dimensionality of data with neural networks,

    G. E. Hinton and R. R. Salakhutdinov, “Re- ducing the dimensionality of data with neural networks,”Science, vol. 313, no. 5786, pp. 504– 507, 2006

    work page 2006

  64. [64]

    Mnist handwritten digit database

    Y. Lecun, C. Cortes, and C. J. Burges, “Mnist handwritten digit database.”ATT Labs [On- line]. Available: http: // yann. lecun. com/ exdb/ mnist, 2010

    work page 2010

  65. [65]

    Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms

    H. Xiao, K. Rasul, and R. Vollgraf, “Fashion- MNIST: a novel image dataset for bench- marking machine learning algorithms,” 2017. https://arxiv.org/abs/1708.07747

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  66. [66]

    Learning multiple layers of features from tiny images,

    A. Krizhevsky, “Learning multiple layers of features from tiny images,” tech. rep., Uni- versity of Toronto, 2009. https://www.cs. toronto.edu/~kriz/cifar.html

    work page 2009

  67. [67]

    Evolution of en- tanglement entropy in one-dimensional sys- tems,

    P. Calabrese and J. Cardy, “Evolution of en- tanglement entropy in one-dimensional sys- tems,”Journal of Statistical Mechanics: The- ory and Experiment, vol. 2005, p. P04010, apr 2005. 17

    work page 2005

  68. [68]

    A short review on entanglement in quantum spin systems,

    J. I. Latorre and A. Riera, “A short review on entanglement in quantum spin systems,” Journal of Physics A: Mathematical and The- oretical, vol. 42, p. 504002, dec 2009

    work page 2009

  69. [69]

    Dynamics of en- tanglement in one-dimensional spin systems,

    L. Amico, A. Osterloh, F. Plastina, R. Fazio, and G. Massimo Palma, “Dynamics of en- tanglement in one-dimensional spin systems,” Phys. Rev. A, vol. 69, p. 022304, Feb 2004. 18

    work page 2004