We settle the long-standing open question of whether parameterized quantum circuits can learn high-dimensional semantic relationships on NISQ hardware and whether quantum phenomena provide measurable advantages. Using IBM's 156-qubit Eagle processor ibm_fez, we show that previous limitations of quantum natural language processing were not fundamental but originated from poor encoding design. All headline results below are obtained on real hardware (no error mitigation, free-tier access, November 2025). Key hardware-validated findings: • Encoding hierarchy: 132× performance gap between faithful Direct Angle Encoding (r = 0.989) and destructive Difference Encoding (r = 0.007) • Genuine quantum learning: +1.61 training effect on hardware, transforming actively anti-correlated random unitaries (r = −0.92) into semantically meaningful manifolds (r = +0.69) with only 21 parameters • Entanglement is required: entangled circuits achieve r = +0.69 while product-state ablation collapses to constant output (r = −0.11), yielding +0.81 entanglement advantage on real hardware • Hardware outperforms simulation: +18% correlation improvement versus noiseless simulator • First empirical scaling law for quantum semantic learning: increasing training data from 12 → 40 pairs improves generalization 4.25× (r = 0.08 → 0.34), projecting parity with classical cosine baseline (r ≈ 0.86) at ~100 training examples We establish three architectural prerequisites for quantum semantic learning on NISQ devices: (1) Sparse Encoding, (2) Ancilla-Based Measurement, and (3) Non-Aliased Angle Scaling [0.1, π−0.1]. This record contains: • Complete manuscript (PDF) • All source code (Qiskit 2.2.3) • Trained parameters (best_theta) • Raw IBM Quantum hardware results (JSON) • Publication-ready figures • 83-pair test set and interactive visualizations All experiments are fully reproducible with an IBM Quantum free-tier account. Related software: https://doi.org/10.5281/zenodo.17728126
We settle the long-standing open question of whether parameterized quantum circuits can learn high-dimensional semantic relationships on NISQ hardware and whether quantum phenomena provide measurable advantages. Using IBM's 156-qubit Eagle processor ibm_fez, we show that previous limitations of quantum natural language processing were not fundamental but originated from poor encoding design. All headline results below are obtained on real hardware (no error mitigation, free-tier access, November 2025). Key hardware-validated findings: • Encoding hierarchy: 132× performance gap between faithful Direct Angle Encoding (r = 0.989) and destructive Difference Encoding (r = 0.007) • Genuine quantum learning: +1.61 training effect on hardware, transforming actively anti-correlated random unitaries (r = −0.92) into semantically meaningful manifolds (r = +0.69) with only 21 parameters • Entanglement is required: entangled circuits achieve r = +0.69 while product-state ablation collapses to constant output (r = −0.11), yielding +0.81 entanglement advantage on real hardware • Hardware outperforms simulation: +18% correlation improvement versus noiseless simulator • First empirical scaling law for quantum semantic learning: increasing training data from 12 → 40 pairs improves generalization 4.25× (r = 0.08 → 0.34), projecting parity with classical cosine baseline (r ≈ 0.86) at ~100 training examples We establish three architectural prerequisites for quantum semantic learning on NISQ devices: (1) Sparse Encoding, (2) Ancilla-Based Measurement, and (3) Non-Aliased Angle Scaling [0.1, π−0.1]. This record contains: • Complete manuscript (PDF) • All source code (Qiskit 2.2.3) • Trained parameters (best_theta) • Raw IBM Quantum hardware results (JSON) • Publication-ready figures • 83-pair test set and interactive visualizations All experiments are fully reproducible with an IBM Quantum free-tier account. Related software: https://doi.org/10.5281/zenodo.17728126
This paper presents the experimental validation and execution of the EIEZ Language (Version 3) on real physical quantum hardware. Deployed across IBM Quantum processors (ibm_fez, ibm_marrakesh, and ibm_torino, featuring 133 to 156 physical qubits), the EIEZ v3 framework introduces hybrid compilation targeting OpenQASM 2.0 and OpenQASM 3.0, adaptive classical control flow (while, repeat-until with rich CondExpr logic), and hardware topology validation (EIEZTopologyError). The test series evaluated 2- and 3-qubit entanglement circuits (Bell State, GHZ State), quantum algorithms (Deutsch), and depth-scaling tests (H×1 to H×10), executed with up to 4,096 shots per job. Key Results: Full Compilation Compatibility: Seamless execution flow from EIEZ source code to OpenQASM and IBM Quantum hardware. High Entanglement Fidelity: Average fidelity of ~95% across Bell State, GHZ, and Deutsch algorithm executions on physical QPUs. Critical Decoherence Point: Empirical observation of quantum interference breakdown at depth H×3 (~50% deviation from the ideal EIEZ simulator), providing physical evidence of criticality and confirming the applicability of the EIE/ZIE adaptive control framework (Patent BR 10 2025 024459 4). Transpiler Depth Recovery: Fidelity restoration to ~99.7% at depth H×10 driven by internal sequence optimization in Qiskit transpilation. Open FAIR Dataset: Release of machine-readable raw measurement counts (JSON/CSV), UTC timestamps, and IBM Job IDs for open verification. Links & Resources: IBM Quantum Cloud Platform: https://quantum.cloud.ibm.com/ Raw Execution Counts (JSON / CSV): https://github.com/joemerson/eiez-lang/blob/main/eiez-lang/Teste_rodados_IBM_/JSON.json Local Simulation Notebooks (Zenodo): https://doi.org/10.5281/zenodo.19078764 EIEZ Compiler Source Code (GitHub): https://github.com/joemerson/eiez-lang Complete EIE/ZIE Theory & Framework (Zenodo): https://doi.org/10.5281/zenodo.19005280 Zenodo DOI: https://doi.org/10.5281/zenodo.19078829
This study investigates classification performance and trainability of variational quantum classifiers on near-term noisy quantum hardware across circuit depth and qubit count regimes within the context of quantum computing and machine learning, an area of growing scientific importance given its implications for hybrid quantum-classical algorithm design and practical NISQ hardware benchmarking for near-term quantum advantage assessment. Using gradient-based parameter optimization of parameterized quantum circuits with hardware noise modeling via density matrix simulation and physical execution on IBM Eagle processors, we examine expressibility of parameterized circuits enabling nonlinear decision boundaries in feature Hilbert space, with trainability constrained by barren plateau gradients at large qubit counts in 96 circuit architectures (4 qubit counts x 4 depths x 6 ansatz families) evaluated across 5 classification datasets with 20 random initialization seeds each drawn from IBM Quantum Eagle (127-qubit) and Falcon (27-qubit) processors with noise mitigation via zero-noise extrapolation. Results indicate that 10-qubit, 4-layer VQC achieves 84.2% classification accuracy on the best benchmark, within 3.8 percentage points of classical SVM, but gradient variance decays exponentially with qubit count (variance halving per 2 added qubits), confirming barren plateau scaling (p < 0.001), with 84.2% accuracy, 3.8 pp gap to classical SVM baseline as the primary quantitative benchmark. Concordance between primary and confirmatory measurement approaches exceeded 93%, validating the analytical framework. These findings contribute empirically to quantum computing and machine learning and carry actionable implications for the design of programs and policies targeting hybrid quantum-classical algorithm design and practical NISQ hardware benchmarking for near-term quantum advantage assessment.
Quantum machine learning on real noisy intermediate-scale quantum (NISQ) hardware has remained largely confined to binary or few-class tasks, limited by the cost of on-hardware training and the underuse of large devices at inference. We present a unified framework that classifies ten-class MNIST end-to-end on a 127-qubit IBM Eagle processor, with three central contributions. First, a two-phase protocol decouples a gradient-based classical optimization of the encoder and readout from a gradient-free optimization of the quantum parameters, removing the parameter-shift gradient cost that makes on-hardware training impractical. Second, we introduce Quantum Multi-Programming to a trained quantum classifier for the first time, packing multiple circuit copies onto one device to deliver parallel inference at no mean-accuracy cost while cutting quantum-processor job submissions proportionally. Third, a controlled comparison shows that on-hardware fine-tuning yields no measurable accuracy gain, motivating a practical NISQ workflow: train on a classical simulator and reserve the hardware for inference only. Benchmarked against a matched-capacity classical network, the quantum module shows no per-parameter accuracy advantage at this scale; we therefore frame the work as a feasibility-and-workflow demonstration for multi-class quantum image classification on current hardware.
Hybrid quantum-classical models represent a crucial step toward leveraging near-term quantum devices for sequential data processing. We present Quantum Recurrent Neural Networks (QRNNs) and Quantum Convolutional Neural Networks (QCNNs) as hybrid quantum language models, reporting the first empirical demonstration of generative language modeling trained and evaluated end-to-end on real quantum hardware. Our architecture combines hardware-optimized parametric quantum circuits with a lightweight classical projection layer, utilizing a multi-sample SPSA strategy to efficiently train quantum parameters despite hardware noise. To characterize the capabilities of these models, we introduce a synthetic dataset designed to isolate syntactic dependencies in a controlled, low-resource environment. Experiments on IBM Quantum processors reveal the critical trade-offs between circuit depth and trainability, demonstrating that while noise remains a significant factor, observable-based readout enables the successful learning of sequential patterns on NISQ devices. These results establish a rigorous engineering baseline for generative quantum natural language processing, validating the feasibility of training complex sequence models on current quantum hardware.
Learning-based Quantum Error Mitigation (QEM) has emerged as a promising approach for reducing noise-induced bias on near-term Noisy Intermediate-Scale Quantum (NISQ) hardware. While simulated noise models enable scalable training of learning-based mitigators, real hardware data more accurately reflect device-specific error mechanisms, but are expensive and sparse. Understanding how these two data regimes interact is therefore critical for the practical deployment of learning-based QEM. In this paper, we present a systematic and empirical analysis on the nature of the data (simulated versus hardware noise) for learning-based QEM. Using a recently developed attention-based graph transformer for small- and large-scale quantum error mitigation, we compare various training regimes, starting from exclusively simulated noise and progressively integrating more measurements from real quantum hardware. Our study spans two distinct circuit families: 1) randomly generated circuits with relatively homogeneous structure; and 2) targeted quantum circuits derived from a specific problem, in this case the one-dimensional Burgers’ equation. Across five-fold cross validation, we analyze how hardware data affects generalization performance as measured by mean absolute error (MAE). Our results demonstrate that the impact of hardware data strongly depends on circuit structure and diversity. For random circuits, simulated data already guarantees lower absolute error, while hardware data yields consistent, but moderate improvements. In contrast, for targeted circuits, like the structured Burgers’ circuits, simulated noise can be less representative of the actual (testing) quantum error, and incorporating hardware data produces substantially larger relative reductions in MAE – despite the intrinsically higher MAE. These findings indicate the value of the nature of the training data in learning-based QEM, which is governed not only by the data’s ability to approximate reality – as expected – but also by how difficult it may be to capture circuit-induced error mechanisms in quantum simulations. In general, this work highlights the urgency of quantum circuit coverage in designing scalable and cost-effective learning-based QEM pipelines for devices in this new NISQ-era.
This paper presents an empirical evaluation of the Qube Engine, a hybrid quantum-classical Variational Quantum Regressor (VQR) implemented on Noisy Intermediate-Scale Quantum (NISQ) systems. The framework integrates classical data encoding via single-qubit rotations, parameterized RY+RZ variational layers, and linear entangling circuits evaluated on 4-qubit quantum registers. Optimization is conducted using classical algorithms (COBYLA) targeting Pauli-Z expectation values and Mean Squared Error (MSE) loss on continuous feature spaces. We evaluate model performance across ideal statevector simulation (AerSimulator) and physical IBM Quantum superconducting processors (ibm_marrakesh, ibm_fez, ibm_kingston). Experimental results demonstrate consistent optimization dynamics and low simulation-to-hardware expectation drift (Δ=0.0188), corresponding to a 98.23% cross-backend output stability score under shallow circuit depths. This work provides a realistic analysis of variational regression dynamics, hardware execution variance, and structural trade-offs for near-term quantum machine learning models on NISQ hardware.