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

Factorial Benchmarking of Exact and Finite-Shot Gradient Resolvability in a Four-Qubit Hybrid Quantum Neural Network

Brandon Shen

Published Aug 11, 2026• 0 citations
Publisher

Abstract

Abstract Gradient resolvability in quantum neural networks can depend simultaneously on circuit structure, objective construction, and finite-shot measurement, yet these interventions are usually benchmarked separately. We use a complete 2³ factorial design to test interactions among a pair-restricted controlled-NOT (CNOT) schedule (E), normalized transverse-field Ising model (TFIM) energy objective (L), and fixed residual shortcut (R) in a four-qubit measure–re-encode–reset hybrid quantum neural network. Across 50 matched initialization clusters and 30 finite-shot replicates per scientific cell, configurations were matched by block count, parameter identity, and total implemented simulator-shot budget. The centered E × L interaction in exact-gradient magnitude was positive and was supported by both mixed-model inference and an initialization-cluster bootstrap; the positive direction was retained in an independent-seed dataset, although its magnitude and depth profile remained uncertain. Finite-shot E × L and E × R interactions met the protocol-derived Wald/Holm rule, but their 1,000-fit nested-bootstrap intervals included zero, and the residual result was estimator-mode sensitive and inactive at D = 1, 2. The L × R-by-depth interaction remained unresolved. These results show that factorial benchmarking can expose composition effects that marginal comparisons miss while separating exact-gradient signal from finite-shot estimator behavior. They do not establish optimization success, hardware advantage, shot savings, or matched physical-resource cost.

DOI
10.21203/rs.3.rs-10646614/v1

Related papers

Gradient usability in few-qubit quantum Neural Networks: A signal-to-noise ratio framework for evaluating mitigation strategies

Brandon Shen, Karena Ling

The barren-plateau problem—the exponential flattening of optimization landscapes—severely limits the trainability of quantum neural networks (QNNs) on near-term hardware. This review examines ansatz-design and complementary mitigation strategies for few-qubit QNNs under finite-shot noisy intermediate-scale quantum (NISQ) constraints. Its primary conceptual contribution is a literature-grounded reframing of gradient-based trainability around pointwise finite-shot gradient signal-to-noise ratio (SNR) while retaining exact-gradient landscape variance as a complementary theoretical diagnostic. This reframing follows because gradient-based optimization acts on finite-shot gradient estimates whose statistical resolvability directly determines whether the optimizer receives a reliably distinguishable update signal, whereas numerical studies restricted to 2–10 qubits generally cannot establish asymptotic exponential or polynomial gradient-decay laws without theoretical support. Estimator SNR therefore addresses the immediate experimental question of whether the update signal is distinguishable from its sampling uncertainty, whereas exact-gradient variance characterizes landscape concentration. The review surveys ansatz design, parameter initialization, cost-function selection, and error management, and develops a conditional design heuristic: problem-inspired ansätze are preferred when the task maps onto circuit families with demonstrated trainability properties; otherwise, shallow hardware-efficient ansätze provide controlled baselines. As a secondary contribution, the review proposes a complete eight-configuration 2 3 factorial framework for evaluating interactions among restricted entanglement, local cost functions, and active residual shortcuts under matched measurement budgets. An exact identity motivates testing the interaction between restricted entanglement and local costs at the exact-gradient-signal level, but the interaction is tested two-sided; all estimator-SNR interactions remain open empirical questions. The framework is presented as an implementable research agenda enabled by the SNR perspective rather than as evidence that any interaction has been established.

Journal of High School Science•Aug 1, 2026•0 citations

Benchmarking Modular Optimization Strategies for Parameterized Quantum Circuits

Carla Cotea, Stefan Balauca, Andreea Arusoaie

Training parameterized quantum circuits requires balancing optimization progress with limited evaluation budgets and the statistical uncertainty of near-term quantum hardware. We present a modular benchmarking framework that separates quantum search-direction estimation from classical parameter-update rules, allowing their interactions and sensitivities to be examined independently. We evaluate a suite of gradient-based, stochastic, and derivative-free optimizers across combinatorial optimization using the Quantum Approximate Optimization Algorithm (QAOA), supervised quantum machine learning using Iris classification and a quantum convolutional neural network (QCNN) for binary MNIST classification, and quantum chemistry using the variational quantum eigensolver (VQE) for molecular hydrogen. By separating objective evaluations from sample-circuit costs under finite-shot simulation and physical execution on a 156-qubit processor, we compare optimizer behavior across workloads with different parameter counts and measurement requirements. Four selected simulator and hardware case studies report terminal and best observed objectives separately, alongside post-training classifier accuracies. The comparisons include selected MaxCut and hydrogen trajectories and illustrate workload-dependent objective changes and evaluation costs, while the simulator benchmark compares both peak and mean performance across three seeds. The hardware runs do not establish an optimizer ranking or isolate a causal effect of device noise.

quant-pharXiv:2610.10254•Oct 8, 2026•0 citations

Factorial Benchmarking of Exact and Finite-Shot Gradient Resolvability in a Four-Qubit Hybrid Quantum Neural Network

Brandon Shen

•Aug 8, 2026•0 citations

Can the Quantum Fisher Effective Dimension Judge a Hybrid Quantum Neural Network?

Md Masum Mia, Amith Khandakar, Mohamed Arselene Ayari

Hybrid quantum neural networks (QNNs) pair a classical feature extractor with a variational quantum circuit, and the effective dimension of the circuit's quantum Fisher information matrix (QFIM) is widely used to assess such models. This scalar is often interpreted as an indicator of trainability or accuracy, although these readings are rarely tested under controlled conditions. We therefore investigate what it actually measures. We define the effective dimension as the participation ratio of the QFIM diagonal, which lies between one and the parameter count, and we evaluate it across two image datasets, four frozen backbones including a random-projection control, four circuit depths on eight qubits, and three seeds. During training, the effective dimension remains well below the parameter count, and its correlation with test accuracy ranges from -0.72 to +0.87, changing sign across seeds in half of the configurations. Increasing depth activates a larger share of the circuit directions, from 0.38 to 0.91 of the parameter count, while the initialization gradient variance remains between 1e-5 and 1e-4, so no barren plateau (BP) appears at eight qubits. At a fixed circuit, the four backbones share an initial effective dimension of 6 to 9, even though CIFAR-10 accuracy ranges from 23.14% to 86.44%. These results show that the effective dimension characterizes circuit geometry rather than model quality. We conclude with a practical protocol that assigns each diagnostic to the question it can answer.

quant-pharXiv:2610.14070•Oct 9, 2026•0 citations

Horizon-Conditioned Proposals for Rare-Event Estimation with Variational and Hybrid Quantum Circuits

Victoria Zhang, Phuong Cao, R. Srikant

Claims about quantum learning models are difficult to audit when the function they approximate is rarely available for comparison. We construct a setting in which it is. For a finite-horizon hitting problem, the optimal importance-sampling proposal is the Doob h-transform, a value function conditioned on the number of steps remaining as well as on the current state. On an enumerable state space, both this target and the exact variance of any proposal derived from it are computable, allowing learning error to be measured rather than estimated. On this testbed, we compare three horizon encodings crossed with two symmetry sectors for Born machines, together with a hybrid dressed circuit and classical regressors, and study sensitivity to initialization, increasingly rare failure regimes, and finite measurement budgets. Within the architectures studied, the statewise-regressive models substantially outperform the Born-machine parameterizations. Across repeated initializations, additive time re-uploading remains the weakest of the tested horizon encodings, while a symmetry-restricted ansatz consistently improves median learning error despite using eight times fewer circuit parameters. Initialization strongly affects absolute performance, yet these broader design trends survive aggregation across seeds. Because both model classes produce normalized proposals, the observed gap reflects differences in parameterization and optimization rather than normalization alone. The two quantum architectures also exhibit distinct sensitivity to finite-shot readout, and across a γ-retuned family spanning p_T = 10^-2 to 10^-8 the classical-to-hybrid gap widens substantially. Together, these results show how an exactly evaluable rare-event benchmark can separate effects of representation, optimization, and measurement that are otherwise easy to conflate in quantum-learning comparisons.

Proceedings of the AAAI Symposium Series•Nov 6, 2026•0 citations

When cheap gradients fail: the measurement cost of attacking quantum classifiers

Bacui Li, Chandra Thapa, Tansu Alpcan +1 more

Adversarial perturbations threaten machine learning classifiers, including variational quantum classifiers. We show that finite quantum measurement statistics (shot noise) act as a built-in defense against gradient-based test-time attacks whose cost scales unfavorably for the attacker. Because every gradient component must be inferred from repeated circuit executions under any unbiased gradient-estimation rule, white-box extraction consumes a dimension-dependent measurement budget that measurement grouping cannot remove in expressive circuits. Under stated assumptions, single-step attacks need at least quadratically many shots in the input dimension $d$, growing as $d^{5/2}$ under norm-concentration scaling, with a sufficient-budget analysis for iterative attacks via stochastic gradient Langevin dynamics. Simulations up to 784 input dimensions validate the law: the realized total budget is the $d^{5/2}$ geometric floor for plateau-mitigated models and grows as $d^{3.00}$ for the tested deep circuits, whose gradient norms decay with dimension absent barren-plateau mitigation; folding the measured gradient norm back in recovers the parameter-free $d^{3/2}$ shot-noise geometry. Against a matched classical baseline whose attack overhead is dimension-independent (the cheap-gradient principle of automatic differentiation), the quantum gradient cost ratio grows empirically as $d^{3.00}$, so the attacker's relative cost diverges as the model scales. Experiments on a 156-qubit IBM processor (ibm_boston, 4-qubit circuits, $d=12$) reproduce the effect: at matched budgets the device attack tracks the ideal within a few percent, with the high-shot gradient faithful to the exact one. The defense operates precisely when the forward map is classically hard to simulate: only then is a white-box attacker denied the simulate-and-backpropagate shortcut and must pay the measurement cost we quantify.

quant-pharXiv:2607.11095•Jul 14, 2026•1 citation
Abstract Gradient resolvability in quantum neural networks can depend simultaneously on circuit structure, objective construction, and finite-shot measurement, yet these interventions are usually benchmarked separately. We use a complete 2³ factorial design to test interactions among a pair-restricted controlled-NOT (CNOT) schedule (E), normalized transverse-field Ising model (TFIM) energy objective (L), and fixed residual shortcut (R) in a four-qubit measure–re-encode–reset hybrid quantum neural network. Across 50 matched initialization clusters and 30 finite-shot replicates per scientific cell, configurations were matched by block count, parameter identity, and total implemented simulator-shot budget. The centered E × L interaction in exact-gradient magnitude was positive and was supported by both mixed-model inference and an initialization-cluster bootstrap; the positive direction was retained in an independent-seed dataset, although its magnitude and depth profile remained uncertain. Finite-shot E × L and E × R interactions met the protocol-derived Wald/Holm rule, but their 1,000-fit nested-bootstrap intervals included zero, and the residual result was estimator-mode sensitive and inactive at D = 1, 2. The L × R-by-depth interaction remained unresolved. These results show that factorial benchmarking can expose composition effects that marginal comparisons miss while separating exact-gradient signal from finite-shot estimator behavior. They do not establish optimization success, hardware advantage, shot savings, or matched physical-resource cost.
The barren-plateau problem—the exponential flattening of optimization landscapes—severely limits the trainability of quantum neural networks (QNNs) on near-term hardware. This review examines ansatz-design and complementary mitigation strategies for few-qubit QNNs under finite-shot noisy intermediate-scale quantum (NISQ) constraints. Its primary conceptual contribution is a literature-grounded reframing of gradient-based trainability around pointwise finite-shot gradient signal-to-noise ratio (SNR) while retaining exact-gradient landscape variance as a complementary theoretical diagnostic. This reframing follows because gradient-based optimization acts on finite-shot gradient estimates whose statistical resolvability directly determines whether the optimizer receives a reliably distinguishable update signal, whereas numerical studies restricted to 2–10 qubits generally cannot establish asymptotic exponential or polynomial gradient-decay laws without theoretical support. Estimator SNR therefore addresses the immediate experimental question of whether the update signal is distinguishable from its sampling uncertainty, whereas exact-gradient variance characterizes landscape concentration. The review surveys ansatz design, parameter initialization, cost-function selection, and error management, and develops a conditional design heuristic: problem-inspired ansätze are preferred when the task maps onto circuit families with demonstrated trainability properties; otherwise, shallow hardware-efficient ansätze provide controlled baselines. As a secondary contribution, the review proposes a complete eight-configuration 2 3 factorial framework for evaluating interactions among restricted entanglement, local cost functions, and active residual shortcuts under matched measurement budgets. An exact identity motivates testing the interaction between restricted entanglement and local costs at the exact-gradient-signal level, but the interaction is tested two-sided; all estimator-SNR interactions remain open empirical questions. The framework is presented as an implementable research agenda enabled by the SNR perspective rather than as evidence that any interaction has been established.
Training parameterized quantum circuits requires balancing optimization progress with limited evaluation budgets and the statistical uncertainty of near-term quantum hardware. We present a modular benchmarking framework that separates quantum search-direction estimation from classical parameter-update rules, allowing their interactions and sensitivities to be examined independently. We evaluate a suite of gradient-based, stochastic, and derivative-free optimizers across combinatorial optimization using the Quantum Approximate Optimization Algorithm (QAOA), supervised quantum machine learning using Iris classification and a quantum convolutional neural network (QCNN) for binary MNIST classification, and quantum chemistry using the variational quantum eigensolver (VQE) for molecular hydrogen. By separating objective evaluations from sample-circuit costs under finite-shot simulation and physical execution on a 156-qubit processor, we compare optimizer behavior across workloads with different parameter counts and measurement requirements. Four selected simulator and hardware case studies report terminal and best observed objectives separately, alongside post-training classifier accuracies. The comparisons include selected MaxCut and hydrogen trajectories and illustrate workload-dependent objective changes and evaluation costs, while the simulator benchmark compares both peak and mean performance across three seeds. The hardware runs do not establish an optimizer ranking or isolate a causal effect of device noise.
Hybrid quantum neural networks (QNNs) pair a classical feature extractor with a variational quantum circuit, and the effective dimension of the circuit's quantum Fisher information matrix (QFIM) is widely used to assess such models. This scalar is often interpreted as an indicator of trainability or accuracy, although these readings are rarely tested under controlled conditions. We therefore investigate what it actually measures. We define the effective dimension as the participation ratio of the QFIM diagonal, which lies between one and the parameter count, and we evaluate it across two image datasets, four frozen backbones including a random-projection control, four circuit depths on eight qubits, and three seeds. During training, the effective dimension remains well below the parameter count, and its correlation with test accuracy ranges from -0.72 to +0.87, changing sign across seeds in half of the configurations. Increasing depth activates a larger share of the circuit directions, from 0.38 to 0.91 of the parameter count, while the initialization gradient variance remains between 1e-5 and 1e-4, so no barren plateau (BP) appears at eight qubits. At a fixed circuit, the four backbones share an initial effective dimension of 6 to 9, even though CIFAR-10 accuracy ranges from 23.14% to 86.44%. These results show that the effective dimension characterizes circuit geometry rather than model quality. We conclude with a practical protocol that assigns each diagnostic to the question it can answer.
Claims about quantum learning models are difficult to audit when the function they approximate is rarely available for comparison. We construct a setting in which it is. For a finite-horizon hitting problem, the optimal importance-sampling proposal is the Doob h-transform, a value function conditioned on the number of steps remaining as well as on the current state. On an enumerable state space, both this target and the exact variance of any proposal derived from it are computable, allowing learning error to be measured rather than estimated. On this testbed, we compare three horizon encodings crossed with two symmetry sectors for Born machines, together with a hybrid dressed circuit and classical regressors, and study sensitivity to initialization, increasingly rare failure regimes, and finite measurement budgets. Within the architectures studied, the statewise-regressive models substantially outperform the Born-machine parameterizations. Across repeated initializations, additive time re-uploading remains the weakest of the tested horizon encodings, while a symmetry-restricted ansatz consistently improves median learning error despite using eight times fewer circuit parameters. Initialization strongly affects absolute performance, yet these broader design trends survive aggregation across seeds. Because both model classes produce normalized proposals, the observed gap reflects differences in parameterization and optimization rather than normalization alone. The two quantum architectures also exhibit distinct sensitivity to finite-shot readout, and across a γ-retuned family spanning p_T = 10^-2 to 10^-8 the classical-to-hybrid gap widens substantially. Together, these results show how an exactly evaluable rare-event benchmark can separate effects of representation, optimization, and measurement that are otherwise easy to conflate in quantum-learning comparisons.
Adversarial perturbations threaten machine learning classifiers, including variational quantum classifiers. We show that finite quantum measurement statistics (shot noise) act as a built-in defense against gradient-based test-time attacks whose cost scales unfavorably for the attacker. Because every gradient component must be inferred from repeated circuit executions under any unbiased gradient-estimation rule, white-box extraction consumes a dimension-dependent measurement budget that measurement grouping cannot remove in expressive circuits. Under stated assumptions, single-step attacks need at least quadratically many shots in the input dimension ddd, growing as d5/2d^{5/2}d5/2 under norm-concentration scaling, with a sufficient-budget analysis for iterative attacks via stochastic gradient Langevin dynamics. Simulations up to 784 input dimensions validate the law: the realized total budget is the d5/2d^{5/2}d5/2 geometric floor for plateau-mitigated models and grows as d3.00d^{3.00}d3.00 for the tested deep circuits, whose gradient norms decay with dimension absent barren-plateau mitigation; folding the measured gradient norm back in recovers the parameter-free d3/2d^{3/2}d3/2 shot-noise geometry. Against a matched classical baseline whose attack overhead is dimension-independent (the cheap-gradient principle of automatic differentiation), the quantum gradient cost ratio grows empirically as d3.00d^{3.00}d3.00, so the attacker's relative cost diverges as the model scales. Experiments on a 156-qubit IBM processor (ibm_boston, 4-qubit circuits, d=12d=12d=12) reproduce the effect: at matched budgets the device attack tracks the ideal within a few percent, with the high-shot gradient faithful to the exact one. The defense operates precisely when the forward map is classically hard to simulate: only then is a white-box attacker denied the simulate-and-backpropagate shortcut and must pay the measurement cost we quantify.