Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2611.00255

Benchmarking Quantum Convolutional Neural Networks and their classical surrogates on long-range strongly correlated models

Tatiana A. Bespalova, Réka Schwengelbeck, Laura Mismetti, Johannes Schachenmayer, Ivano Tavernelli, Francesco Tacchino

quant-phSubmitted Oct 30, 2026• Announced Nov 3, 2026• 0 citations
arXivHTMLPDF

Abstract

We benchmark quantum convolutional neural networks (QCNNs) and their classical surrogates for phase differentiation in one-dimensional strongly correlated systems with beyond-nearest-neighbour terms. Using an anisotropic Heisenberg spin chain and an extended $t$-$t'$ Fermi-Hubbard model as representative spin and fermionic testbeds, we train QCNN classifiers on labelled ground-state datasets and map out finite-size phase diagrams. In hybrid setups, we employ Pauli backpropagation to express the QCNN measurement as a weighted sum of Pauli-string expectation values to be evaluated directly on input states, and study how truncation parameters control the trade-off between approximation accuracy and the size of the resulting Pauli expansion. Robust convergence is achieved for the Heisenberg model under aggressive truncation, whereas the Fermi-Hubbard model requires substantially looser thresholds, resulting in a sharp growth in retained Pauli strings and evaluation cost. Our results contribute to charting where hybrid quantum-classical training is viable for learning strongly correlated quantum states, and provide practical guidance for choosing truncation regimes in Pauli backpropagation-based surrogates.

AI Summary

The paper asks when truncated Pauli-backpropagation surrogates can replace quantum convolutional neural network (QCNN) execution and training for phase classification of strongly correlated ground states. It benchmarks an anisotropic Heisenberg chain with next-nearest-neighbour exchange and a half-filled $t$–$t'$ Fermi–Hubbard chain, using exact-diagonalisation inputs and noiseless simulations at $N=8,12$ qubits (four and six Hubbard sites). QCNNs are trained on binary phase labels, typically using 27 ground states, with additional five-state training tests. The resulting maps qualitatively reproduce selected phase regions, although the six-site Hubbard system requires finite-size-adjusted labels rather than thermodynamic-limit labels. Backpropagating the output observable into a truncated Pauli expansion establishes accuracy–cost trade-offs through output deviations, classification disagreements, and retained-string counts. At fixed trained parameters, Pauli-weight cutoffs around six yield near-minimal classification disagreement in the studied settings. End-to-end surrogate optimisation is more model-dependent. For Heisenberg inputs, maximum Pauli weight four, frequency cutoff 40, and coefficient threshold $10^{-4}$ support convergence at both sizes; the more aggressive combination $(3,30,10^{-3})$ works at eight qubits but stalls at twelve. Hubbard training requires looser truncation and larger expansions. Jordan–Wigner nonlocality is proposed as a possible explanation, without an operator-level analysis. The study does not establish large-system scaling or include state-preparation errors or finite-shot training. Endpoint agreement is explicitly distinguished from trainability. Larger-system tests, alternative fermionic encodings, and interpretation of learned observables are left for future work.

Little sign of AI-generated textIncrementalSmall numerics onlyMissing comparison

Central claim may not hold: The supplied results do not explicitly document a successful relaxed-truncation Hubbard surrogate run with its convergence and Pauli count, and the table’s nearly untruncated rows explicitly are not surrogate-training runs. The paper does not report actual-circuit held-out performance at surrogate-trained parameters, leaving preservation of the QCNN classifier unverified by the reported surrogate training losses. The shot-cost formula equates a coefficient-discard threshold with measurement precision without accounting for coefficient-weighted accumulation of expectation-estimation errors.

Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.

Comments
13 pages, 11 figures

Related papers

Quantum Convolutional Neural Networks are Effectively Classically Simulable

Pablo Bermejo, Paolo Braccia, Manuel S. Rudolph +3 more

Quantum convolutional neural networks (QCNNs) are widely regarded as a promising model for quantum machine learning (QML). In this work, we analyze the most widely used variants of these models (i.e., tracing out- and measurement-based QCNNs), and we relate their heuristic success to two facts. First, that when randomly initialized, they can only operate on the information encoded in low-bodyness measurements of their input states. And second, that they are commonly benchmarked on “locally easy” datasets whose states are precisely classifiable by the information encoded in these low-bodyness observables subspace. From these insights, we argue that the QCNN’s action on this subspace should be efficiently classically simulable. Indeed, we construct and train a purely classical QCNN surrogate—based on low-bodyness Pauli propagation, tensor networks, and classical shadows—that matches or outperforms standard QCNNs on all benchmark datasets and on up-to 1024 qubits, thereby empirically realizing our simulability claims. Our results can then be understood as highlighting a deeper symptom of QML: Models could only be showing heuristic success because they are benchmarked on simple problems, for which their action can be classically simulated. This insight points to the fact that nontrivial datasets are a truly necessary ingredient for moving forward with QML. To finish, we discuss how our results can be extrapolated to classically simulate other architectures.

arXiv:2408.12739•PRX Quantum•Feb 19, 2026•133 citations

A Comparative Study of Encoding Strategies for Quantum Convolutional Neural Networks

Xingyun Feng

Quantum convolutional neural networks (QCNNs) offer a promising architecture for near-term quantum machine learning by combining hierarchical feature extraction with modest parameter growth. However, any QCNN operating on classical data must rely on an encoding scheme to embed inputs into quantum states, and this choice can dominate both performance and resource requirements. This work presents an implementation-level comparison of three representative encodings -- Angle, Amplitude, and a Hybrid phase/angle scheme -- for QCNNs under depolarizing noise. We develop a fully differentiable PyTorch--Qiskit pipeline with a custom autograd bridge, batched parameter-shift gradients, and shot scheduling, and use it to train QCNNs on downsampled binary variants of MNIST and Fashion-MNIST at $4\times 4$ and $8\times 8$ resolutions. Our experiments reveal regime-dependent trade-offs. On aggressively downsampled $4\times 4$ inputs, Angle encoding attains higher accuracy and remains comparatively robust as noise increases, while the Hybrid encoder trails and exhibits non-monotonic trends. At $8\times 8$, the Hybrid scheme can overtake Angle under moderate noise, suggesting that mixed phase/angle encoders benefit from additional feature bandwidth. Amplitude-encoded QCNNs are sparsely represented in the downsampled grids but achieve strong performance in lightweight and full-resolution configurations, where training dynamics closely resemble classical convergence. Taken together, these results provide practical guidance for choosing QCNN encoders under joint constraints of resolution, noise strength, and simulation budget.

quant-pharXiv:2512.12512•Dec 16, 2025•1 citation

Detecting Quantum Critical Points of Correlated Systems by Quantum Convolutional Neural Network Using Data from Variational Quantum Eigensolver

Nathaniel Wrobel, Anshumitra Baul, Ka-Ming Tam +1 more

Machine learning has been applied to a wide variety of models, from classical statistical mechanics to quantum strongly correlated systems, for classifying phase transitions. The recently proposed quantum convolutional neural network (QCNN) provides a new framework for using quantum circuits instead of classical neural networks as the backbone of classification methods. We present the results from training the QCNN by the wavefunctions of the variational quantum eigensolver for the one-dimensional transverse field Ising model (TFIM). We demonstrate that the QCNN identifies wavefunctions corresponding to the paramagnetic and ferromagnetic phases of the TFIM with reasonable accuracy. The QCNN can be trained to predict the corresponding ‘phase’ of wavefunctions around the putative quantum critical point even though it is trained by wavefunctions far away. The paper provides a basis for exploiting the QCNN to identify the quantum critical point.

arXiv:2111.05076•Quantum Reports•Dec 8, 2022•9 citations

Realizing quantum convolutional neural networks on a superconducting quantum processor to recognize quantum phases

J. Herrmann, Sergi Masot-Llima, Ants Remm +15 more

Quantum computing crucially relies on the ability to efficiently characterize the quantum states output by quantum hardware. Conventional methods which probe these states through direct measurements and classically computed correlations become computationally expensive when increasing the system size. Quantum neural networks tailored to recognize specific features of quantum states by combining unitary operations, measurements and feedforward promise to require fewer measurements and to tolerate errors. Here, we realize a quantum convolutional neural network (QCNN) on a 7-qubit superconducting quantum processor to identify symmetry-protected topological (SPT) phases of a spin model characterized by a non-zero string order parameter. We benchmark the performance of the QCNN based on approximate ground states of a family of cluster-Ising Hamiltonians which we prepare using a hardware-efficient, low-depth state preparation circuit. We find that, despite being composed of finite-fidelity gates itself, the QCNN recognizes the topological phase with higher fidelity than direct measurements of the string order parameter for the prepared states.

arXiv:2109.05909•Nature Communications•Jul 16, 2022•135 citations

A novel quantum convolutional neural network framework for quantum-enhanced classification of pixelated colour images

Chisomo Daka, Somnath Bhattacharyya

As image datasets grow in scale and complexity, classical convolutional neural networks (CNNs) increasingly face limitations in computational efficiency, scalability, and generalisation in low-data regimes. Quantum machine learning (QML) offers a promising alternative, with quantum convolutional neural networks (QCNNs) exploiting quantum parallelism and entanglement for feature extraction. This paper introduces the Novel Quantum Convolutional Neural Network (No-QCNN), which differs from recent hybrid QCNN/QCQ-CNN architectures by employing an end-to-end quantum convolution-pooling pipeline without inserting an intermediate classical CNN block. No-QCNN is a hybrid quantum-classical model that employs a variational quantum classifier (VQC) optimised via the COBYLA algorithm and is designed for binary and multiclass classification of low-resolution colour images on near-term quantum devices. A key innovation is a problem-specific quantum feature map that pre-processes each image into a structured three-dimensional block-matrix representation, jointly encoding pixel colour (R, G, B) and spatial position before mapping this information into a hierarchical ZZFeatureMap. This encoding captures spatial-chromatic correlations at shallow circuit depth, making it compatible with noisy intermediate-scale quantum (NISQ) constraints. The model is implemented using IBM-Qiskit on a local quantum simulator and benchmarked against a classical CNN. For a six-class classification task on a 50-image dataset, No-QCNN achieves a validation accuracy of 82.05%, substantially outperforming the classical CNN's 40.00%, indicating improved generalisation and reduced overfitting in low-data, multiclass settings. Conversely, for a simpler binary classification task, the classical CNN achieves perfect validation accuracy of 100%, surpassing the No-QCNN's 89.7%. We further observe that No-QCNN performance declines with increasing dataset size and training time due to limited circuit expressibility, data-encoding overhead, and finite-sampling noise inherent to NISQ models. Overall, these results position No-QCNN as a complementary framework best suited to low-data, correlation-rich classification tasks, defining a realistic niche for quantum-enhanced artificial vision and quantum perception in the NISQ era.

Scientific Reports•Mar 27, 2026•4 citations

Error-tolerant quantum convolutional neural networks for symmetry-protected topological phases

Petr Zapletal, Nathan A. McMahon, Michael J. Hartmann

The analysis of noisy quantum states prepared on current quantum computers is getting beyond the capabilities of classical computing. Quantum neural networks based on parametrized quantum circuits, measurements and feed-forward can process large amounts of quantum data to reduce measurement and computational costs of detecting nonlocal quantum correlations. The tolerance of errors due to decoherence and gate infidelities is a key requirement for the application of quantum neural networks on near-term quantum computers. Here we construct quantum convolutional neural networks (QCNNs) that can, in the presence of incoherent errors, recognize different symmetry-protected topological phases of generalized cluster-Ising Hamiltonians from one another as well as from topologically trivial phases. Using matrix product state simulations, we show that the QCNN output is robust against symmetry-breaking errors below a threshold error probability and against symmetry-preserving errors provided the error channel is invertible. This is in contrast to string order parameters and the output of previously designed QCNNs, which vanish in the presence of any symmetry-breaking errors. To facilitate the implementation of the QCNNs on near-term quantum computers, the QCNN circuits can be shortened from logarithmic to constant depth in system size by performing a large part of the computation in classical postprocessing. These constant-depth QCNNs reduce sample complexity exponentially with system size in comparison to the direct sampling using local Pauli measurements. Published by the American Physical Society 2024

arXiv:2307.03711•Physical Review Research•Jul 26, 2024•13 citations
We benchmark quantum convolutional neural networks (QCNNs) and their classical surrogates for phase differentiation in one-dimensional strongly correlated systems with beyond-nearest-neighbour terms. Using an anisotropic Heisenberg spin chain and an extended ttt-t′t't′ Fermi-Hubbard model as representative spin and fermionic testbeds, we train QCNN classifiers on labelled ground-state datasets and map out finite-size phase diagrams. In hybrid setups, we employ Pauli backpropagation to express the QCNN measurement as a weighted sum of Pauli-string expectation values to be evaluated directly on input states, and study how truncation parameters control the trade-off between approximation accuracy and the size of the resulting Pauli expansion. Robust convergence is achieved for the Heisenberg model under aggressive truncation, whereas the Fermi-Hubbard model requires substantially looser thresholds, resulting in a sharp growth in retained Pauli strings and evaluation cost. Our results contribute to charting where hybrid quantum-classical training is viable for learning strongly correlated quantum states, and provide practical guidance for choosing truncation regimes in Pauli backpropagation-based surrogates.
The paper asks when truncated Pauli-backpropagation surrogates can replace quantum convolutional neural network (QCNN) execution and training for phase classification of strongly correlated ground states. It benchmarks an anisotropic Heisenberg chain with next-nearest-neighbour exchange and a half-filled ttt–t′t't′ Fermi–Hubbard chain, using exact-diagonalisation inputs and noiseless simulations at N=8,12N=8,12N=8,12 qubits (four and six Hubbard sites).
QCNNs are trained on binary phase labels, typically using 27 ground states, with additional five-state training tests. The resulting maps qualitatively reproduce selected phase regions, although the six-site Hubbard system requires finite-size-adjusted labels rather than thermodynamic-limit labels. Backpropagating the output observable into a truncated Pauli expansion establishes accuracy–cost trade-offs through output deviations, classification disagreements, and retained-string counts. At fixed trained parameters, Pauli-weight cutoffs around six yield near-minimal classification disagreement in the studied settings.
End-to-end surrogate optimisation is more model-dependent. For Heisenberg inputs, maximum Pauli weight four, frequency cutoff 40, and coefficient threshold 10−410^{-4}10−4 support convergence at both sizes; the more aggressive combination (3,30,10−3)(3,30,10^{-3})(3,30,10−3) works at eight qubits but stalls at twelve. Hubbard training requires looser truncation and larger expansions. Jordan–Wigner nonlocality is proposed as a possible explanation, without an operator-level analysis.
The study does not establish large-system scaling or include state-preparation errors or finite-shot training. Endpoint agreement is explicitly distinguished from trainability. Larger-system tests, alternative fermionic encodings, and interpretation of learned observables are left for future work.
Quantum convolutional neural networks (QCNNs) are widely regarded as a promising model for quantum machine learning (QML). In this work, we analyze the most widely used variants of these models (i.e., tracing out- and measurement-based QCNNs), and we relate their heuristic success to two facts. First, that when randomly initialized, they can only operate on the information encoded in low-bodyness measurements of their input states. And second, that they are commonly benchmarked on “locally easy” datasets whose states are precisely classifiable by the information encoded in these low-bodyness observables subspace. From these insights, we argue that the QCNN’s action on this subspace should be efficiently classically simulable. Indeed, we construct and train a purely classical QCNN surrogate—based on low-bodyness Pauli propagation, tensor networks, and classical shadows—that matches or outperforms standard QCNNs on all benchmark datasets and on up-to 1024 qubits, thereby empirically realizing our simulability claims. Our results can then be understood as highlighting a deeper symptom of QML: Models could only be showing heuristic success because they are benchmarked on simple problems, for which their action can be classically simulated. This insight points to the fact that nontrivial datasets are a truly necessary ingredient for moving forward with QML. To finish, we discuss how our results can be extrapolated to classically simulate other architectures.
Quantum convolutional neural networks (QCNNs) offer a promising architecture for near-term quantum machine learning by combining hierarchical feature extraction with modest parameter growth. However, any QCNN operating on classical data must rely on an encoding scheme to embed inputs into quantum states, and this choice can dominate both performance and resource requirements. This work presents an implementation-level comparison of three representative encodings -- Angle, Amplitude, and a Hybrid phase/angle scheme -- for QCNNs under depolarizing noise. We develop a fully differentiable PyTorch--Qiskit pipeline with a custom autograd bridge, batched parameter-shift gradients, and shot scheduling, and use it to train QCNNs on downsampled binary variants of MNIST and Fashion-MNIST at 4×44\times 44×4 and 8×88\times 88×8 resolutions. Our experiments reveal regime-dependent trade-offs. On aggressively downsampled 4×44\times 44×4 inputs, Angle encoding attains higher accuracy and remains comparatively robust as noise increases, while the Hybrid encoder trails and exhibits non-monotonic trends. At 8×88\times 88×8, the Hybrid scheme can overtake Angle under moderate noise, suggesting that mixed phase/angle encoders benefit from additional feature bandwidth. Amplitude-encoded QCNNs are sparsely represented in the downsampled grids but achieve strong performance in lightweight and full-resolution configurations, where training dynamics closely resemble classical convergence. Taken together, these results provide practical guidance for choosing QCNN encoders under joint constraints of resolution, noise strength, and simulation budget.
Machine learning has been applied to a wide variety of models, from classical statistical mechanics to quantum strongly correlated systems, for classifying phase transitions. The recently proposed quantum convolutional neural network (QCNN) provides a new framework for using quantum circuits instead of classical neural networks as the backbone of classification methods. We present the results from training the QCNN by the wavefunctions of the variational quantum eigensolver for the one-dimensional transverse field Ising model (TFIM). We demonstrate that the QCNN identifies wavefunctions corresponding to the paramagnetic and ferromagnetic phases of the TFIM with reasonable accuracy. The QCNN can be trained to predict the corresponding ‘phase’ of wavefunctions around the putative quantum critical point even though it is trained by wavefunctions far away. The paper provides a basis for exploiting the QCNN to identify the quantum critical point.
Quantum computing crucially relies on the ability to efficiently characterize the quantum states output by quantum hardware. Conventional methods which probe these states through direct measurements and classically computed correlations become computationally expensive when increasing the system size. Quantum neural networks tailored to recognize specific features of quantum states by combining unitary operations, measurements and feedforward promise to require fewer measurements and to tolerate errors. Here, we realize a quantum convolutional neural network (QCNN) on a 7-qubit superconducting quantum processor to identify symmetry-protected topological (SPT) phases of a spin model characterized by a non-zero string order parameter. We benchmark the performance of the QCNN based on approximate ground states of a family of cluster-Ising Hamiltonians which we prepare using a hardware-efficient, low-depth state preparation circuit. We find that, despite being composed of finite-fidelity gates itself, the QCNN recognizes the topological phase with higher fidelity than direct measurements of the string order parameter for the prepared states.
As image datasets grow in scale and complexity, classical convolutional neural networks (CNNs) increasingly face limitations in computational efficiency, scalability, and generalisation in low-data regimes. Quantum machine learning (QML) offers a promising alternative, with quantum convolutional neural networks (QCNNs) exploiting quantum parallelism and entanglement for feature extraction. This paper introduces the Novel Quantum Convolutional Neural Network (No-QCNN), which differs from recent hybrid QCNN/QCQ-CNN architectures by employing an end-to-end quantum convolution-pooling pipeline without inserting an intermediate classical CNN block. No-QCNN is a hybrid quantum-classical model that employs a variational quantum classifier (VQC) optimised via the COBYLA algorithm and is designed for binary and multiclass classification of low-resolution colour images on near-term quantum devices. A key innovation is a problem-specific quantum feature map that pre-processes each image into a structured three-dimensional block-matrix representation, jointly encoding pixel colour (R, G, B) and spatial position before mapping this information into a hierarchical ZZFeatureMap. This encoding captures spatial-chromatic correlations at shallow circuit depth, making it compatible with noisy intermediate-scale quantum (NISQ) constraints. The model is implemented using IBM-Qiskit on a local quantum simulator and benchmarked against a classical CNN. For a six-class classification task on a 50-image dataset, No-QCNN achieves a validation accuracy of 82.05%, substantially outperforming the classical CNN's 40.00%, indicating improved generalisation and reduced overfitting in low-data, multiclass settings. Conversely, for a simpler binary classification task, the classical CNN achieves perfect validation accuracy of 100%, surpassing the No-QCNN's 89.7%. We further observe that No-QCNN performance declines with increasing dataset size and training time due to limited circuit expressibility, data-encoding overhead, and finite-sampling noise inherent to NISQ models. Overall, these results position No-QCNN as a complementary framework best suited to low-data, correlation-rich classification tasks, defining a realistic niche for quantum-enhanced artificial vision and quantum perception in the NISQ era.
The analysis of noisy quantum states prepared on current quantum computers is getting beyond the capabilities of classical computing. Quantum neural networks based on parametrized quantum circuits, measurements and feed-forward can process large amounts of quantum data to reduce measurement and computational costs of detecting nonlocal quantum correlations. The tolerance of errors due to decoherence and gate infidelities is a key requirement for the application of quantum neural networks on near-term quantum computers. Here we construct quantum convolutional neural networks (QCNNs) that can, in the presence of incoherent errors, recognize different symmetry-protected topological phases of generalized cluster-Ising Hamiltonians from one another as well as from topologically trivial phases. Using matrix product state simulations, we show that the QCNN output is robust against symmetry-breaking errors below a threshold error probability and against symmetry-preserving errors provided the error channel is invertible. This is in contrast to string order parameters and the output of previously designed QCNNs, which vanish in the presence of any symmetry-breaking errors. To facilitate the implementation of the QCNNs on near-term quantum computers, the QCNN circuits can be shortened from logarithmic to constant depth in system size by performing a large part of the computation in classical postprocessing. These constant-depth QCNNs reduce sample complexity exponentially with system size in comparison to the direct sampling using local Pauli measurements. Published by the American Physical Society 2024