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

Advanced Quantum Technologies

Feature Fusion‐Based Hybrid Quantum‐Classical Graph Residual Neural Network

Kairan Zhang, Desheng Kong, Kangning An, Mingyang Yu, Ji Du, Yulong Fu, Jing Xu, Donglin Wang

Published Feb 1, 2026• 0 citations
Publisher

Abstract

ABSTRACT In the node classification task on graphs, mainstream graph neural networks often adopt the same neighborhood aggregation strategy for all nodes. This approach results in biased learning toward features from the majority nodes while devaluing those from the minority nodes under class imbalance. In particular, for marginal nodes, limited neighboring information can severely impact classification performance. To address this challenge, this paper proposes a feature fusion‐based hybrid quantum‐classical graph residual neural network (QGRNN). Leveraging the nonlinear expressive capacity of qubits in modeling complex feature interactions, the model innovatively integrates a structure‐driven node selection mechanism with a quantum feature enhancement module, while also dynamically fusing classical features and Hamiltonian expectation values through a gated residual fusion mechanism to compensate for representational deficiencies of marginal nodes overlooked by classical methods. Experimental results show that QGRNN consistently outperforms baselines across a range of node classification tasks. In binary and ternary classification settings, it exhibits strong discriminative capability and robustness, especially maintaining high accuracy under severe class imbalance. In addition, QGRNN also demonstrates strong generality in other tasks, in recommendation scenarios, achieving an average improvement of 36.3% on NDCG@5, 44.3% on Recall@5, and 21.9% on Recall@10 compared to the baseline.

DOI
10.1002/qute.202500625

Related papers

Feature prediction in quantum graph recurrent neural networks with applications in information hiding

Jawaher Kaldari, Saif M. Al-Kuwari

Abstract Graphs are fundamental structures for representing complex, nonlinear structured data across various domains, including social networks and quantum systems. Quantum graph recurrent neural networks (QGRNNs) have been proposed to model quantum dynamics in graph-based quantum systems, but their applicability to classical data remains an open problem. In this paper, we leverage QGRNNs to process classical graph-structured data. In particular, we demonstrate how QGRNN can reconstruct node features in classical datasets. Our results show that QGRNN achieves high feature reconstruction accuracy, resulting in near-perfect classification. Furthermore, we propose an information hiding technique based on our QGRNN, where messages are embedded into a graph and then retrieved under certain conditions. We assess retrieval accuracy for different dictionary sizes and message lengths, showing that QGRNN maintains high retrieval accuracy, with minor degradation as complexity increases. These findings demonstrate the scalability and robustness of QGRNNs for both classical data processing and secure information hiding, paving the way toward quantum-enhanced feature extraction, privacy-preserving computations, and quantum steganography.

arXiv:2506.23144•Quantum Information Processing•Feb 6, 2026•0 citations

Integrating Quantum Feature Maps into Graph Neural Networks for Robust Visual Localization

Emma Bennett, Ricky D. Massaro, Filiz Bunyak

Robust aerial localization in GNSS-denied environments is a critical challenge for autonomous navigation. This paper introduces a hybrid Quantum-Classical Graph Neural Network (QGNN) framework for visual localization, treating the task as an approximate subgraph matching problem robust to structural perturbations. We propose and evaluate two architectures for integrating quantum feature maps into a classical graph convolution backbone: a hardware-efficient global projection, where a single quantum circuit processes the pooled graph embedding, and a structurally expressive local projection, where quantum circuits are applied to individual node embeddings prior to aggregation. The models are trained via triplet loss on geospatial graphs constructed from the Massachusetts Buildings dataset. We demonstrate that while global methods are computationally tractable on current IBM Quantum hardware, the local quantum-classical hybrid significantly outperforms classical baselines in high-noise regimes. Specifically, the local quantum architecture improves Recall@1 by up to 20 percentage points under severe structural perturbations (e.g., node deletions and additions), suggesting that node-level quantum feature mapping provides improved regularization against topological noise.

International Joint Conference on Neural Networks (IJCNN)•Jun 1, 2026•0 citations

Practical Quantum-Classical Feature Fusion for complex data Classification

Azadeh Alavi, Fatemeh Kouchmeshki, Abdolrahman Alavi

Hybrid quantum and classical learning aims to couple quantum feature maps with the robustness of classical neural networks, yet most architectures treat the quantum circuit as an isolated feature extractor and merge its measurements with classical representations by direct concatenation. This neglects that the quantum and classical branches constitute distinct computational modalities and limits reliable performance on complex, high dimensional tabular and semi structured data, including remote sensing, environmental monitoring, and medical diagnostics. We present a multimodal formulation of hybrid learning and propose a cross attention mid fusion architecture in which a classical representation queries quantum derived feature tokens through an attention block with residual connectivity. The quantum branch is kept within practical NISQ budgets and uses up to nine qubits. We evaluate on Wine, Breast Cancer, Forest CoverType, FashionMNIST, and SteelPlatesFaults, comparing a quantum only model, a classical baseline, residual hybrid models, and the proposed mid fusion model under a consistent protocol. Pure quantum and standard hybrid designs underperform due to measurement induced information loss, while cross attention mid fusion is consistently competitive and improves performance on the more complex datasets in most cases. These findings suggest that quantum derived information becomes most valuable when integrated through principled multimodal fusion rather than used in isolation or loosely appended to classical features.

arXiv:2512.19180•Dec 22, 2025•1 citation

Parallel structure of hybrid quantum–classical neural networks for image classification

Zuyu Xu, Yuanming Hu, Tao Yang +8 more

Abstract Hybrid quantum-classical neural networks (QCNNs) integrate principles from quantum computing principle and classical neural networks, offering a novel computational approach for image classification tasks. However, current QCNNs with sequential structures encounter limitations in accuracy and robustness, especially when dealing with tasks involving numerous classes. In this study, we propose a novel solution - the hybrid Parallel Quantum Classical Neural Network (PQCNN) - for image classification tasks. This architecture seamlessly integrates the parallel processing capabilities of quantum computing with the hierarchical feature extraction abilities of classical neural networks, aiming to overcome the constraints of conventional sequential structures in multi-class classification tasks. Extensive experimentation demonstrates the superiority of PQCNN over traditional concatenative structures in binary classification datasets, displaying heightened accuracy and robustness against noise. Particularly noteworthy is PQCNN's significantly improved accuracy on datasets with 5 and 10 classes. These findings underscore the transformative potential of the PQCNN architecture as an advanced solution for enhancing the performance of quantum-classical-based classifiers, particularly in the domain of image classification.

Quantum Information Processing•Jun 23, 2025•13 citations

Feature-Adaptive Fusion in Hybrid Quantum-Classical Neural Networks for Robust Biomedical Image Classification

Yan-Yan Hou, Jian Li, Chongqiang Ye +3 more

Hybrid quantum-classical neural networks provide a promising approach for incorporating quantum circuits into machine learning in the noisy intermediate-scale quantum regime. However, existing hybrid models often rely on fixed or globally shared fusion strategies, which may limit their ability to exploit complementary information carried by quantum branches, especially under distribution shifts. In this work, we propose a Feature-Adaptive Fusion Hybrid Quantum-Classical Neural Network (FAF-HQNN) for biomedical image classification. The model combines a classical deep feature encoder with a variational quantum circuit (VQC) and introduces a feature-adaptive fusion mechanism to dynamically weight classical and quantum predictions. We evaluate FAF-HQNN on two MedMNIST benchmarks, PathMNIST and BloodMNIST, under clean and corrupted test conditions. FAF-HQNN achieves the strongest overall performance on clean data among the compared methods and shows improved robustness under Gaussian, salt-and-pepper, and Poisson corruptions. Further analysis of circuit layout, measurement basis, and depth shows that even shallow variational quantum circuits can provide useful complementary information. These results demonstrate that feature-adaptive fusion is an effective strategy for improving accuracy and robustness in hybrid quantum-classical models for biomedical image classification.

quant-pharXiv:2609.15267•Quantum Machine Intelligence•Sep 15, 2026•0 citations

Hybrid Quantum-Classical Graph Neural Networks for Solving Unit Commitment Problem

Xuanzi Ma, Junchao Cui, Wenxia Wang +3 more

The Unit Commitment Problem (UCP) is a core optimization task for ensuring secure and economically efficient operation of power grids. However, existing methodologies face two major limitations: first, there is a lack of a general and efficient way to represent the complex relationships between UCP variables and constraints; second, exponential parameter growth with problem scale impedes characterization of complex dependencies in UCP, thereby limiting predictive accuracy. To address these issues, we propose a hybrid quantum-classical graph neural network framework (HQGNN) that establishes a novel paradigm for UCP by: (1) modeling UCP as a unified bipartite graph format, and (2) integrating quantum feature extraction modules to reduce the parameter size and enhance classical GNNs’ capability in capturing high-dimensional structural characteristics. Extensive experiments on IEEE and RTS benchmark datasets demonstrate that HQGNN achieves improvements in UCP solution accuracy while validating the effectiveness of quantum modules in enhancing graph-structured feature learning. The quantum module in HQGNN is proven to be the key enabler of performance gains. Its integration cuts parameters by 16.8% and reduces mean squared error (MSE) in optimal solution prediction by 24.4% across all benchmarks, while achieving full-task superiority on the IEEE-9 dataset.These results substantiate the proposed method’s capacity to simultaneously enhance solution quality and computational efficiency through quantum-enhanced graph representation learning.

International Joint Conference on Neural Networks (IJCNN)•Jun 1, 2026•0 citations
ABSTRACT In the node classification task on graphs, mainstream graph neural networks often adopt the same neighborhood aggregation strategy for all nodes. This approach results in biased learning toward features from the majority nodes while devaluing those from the minority nodes under class imbalance. In particular, for marginal nodes, limited neighboring information can severely impact classification performance. To address this challenge, this paper proposes a feature fusion‐based hybrid quantum‐classical graph residual neural network (QGRNN). Leveraging the nonlinear expressive capacity of qubits in modeling complex feature interactions, the model innovatively integrates a structure‐driven node selection mechanism with a quantum feature enhancement module, while also dynamically fusing classical features and Hamiltonian expectation values through a gated residual fusion mechanism to compensate for representational deficiencies of marginal nodes overlooked by classical methods. Experimental results show that QGRNN consistently outperforms baselines across a range of node classification tasks. In binary and ternary classification settings, it exhibits strong discriminative capability and robustness, especially maintaining high accuracy under severe class imbalance. In addition, QGRNN also demonstrates strong generality in other tasks, in recommendation scenarios, achieving an average improvement of 36.3% on NDCG@5, 44.3% on Recall@5, and 21.9% on Recall@10 compared to the baseline.
Abstract Graphs are fundamental structures for representing complex, nonlinear structured data across various domains, including social networks and quantum systems. Quantum graph recurrent neural networks (QGRNNs) have been proposed to model quantum dynamics in graph-based quantum systems, but their applicability to classical data remains an open problem. In this paper, we leverage QGRNNs to process classical graph-structured data. In particular, we demonstrate how QGRNN can reconstruct node features in classical datasets. Our results show that QGRNN achieves high feature reconstruction accuracy, resulting in near-perfect classification. Furthermore, we propose an information hiding technique based on our QGRNN, where messages are embedded into a graph and then retrieved under certain conditions. We assess retrieval accuracy for different dictionary sizes and message lengths, showing that QGRNN maintains high retrieval accuracy, with minor degradation as complexity increases. These findings demonstrate the scalability and robustness of QGRNNs for both classical data processing and secure information hiding, paving the way toward quantum-enhanced feature extraction, privacy-preserving computations, and quantum steganography.
Robust aerial localization in GNSS-denied environments is a critical challenge for autonomous navigation. This paper introduces a hybrid Quantum-Classical Graph Neural Network (QGNN) framework for visual localization, treating the task as an approximate subgraph matching problem robust to structural perturbations. We propose and evaluate two architectures for integrating quantum feature maps into a classical graph convolution backbone: a hardware-efficient global projection, where a single quantum circuit processes the pooled graph embedding, and a structurally expressive local projection, where quantum circuits are applied to individual node embeddings prior to aggregation. The models are trained via triplet loss on geospatial graphs constructed from the Massachusetts Buildings dataset. We demonstrate that while global methods are computationally tractable on current IBM Quantum hardware, the local quantum-classical hybrid significantly outperforms classical baselines in high-noise regimes. Specifically, the local quantum architecture improves Recall@1 by up to 20 percentage points under severe structural perturbations (e.g., node deletions and additions), suggesting that node-level quantum feature mapping provides improved regularization against topological noise.
Hybrid quantum and classical learning aims to couple quantum feature maps with the robustness of classical neural networks, yet most architectures treat the quantum circuit as an isolated feature extractor and merge its measurements with classical representations by direct concatenation. This neglects that the quantum and classical branches constitute distinct computational modalities and limits reliable performance on complex, high dimensional tabular and semi structured data, including remote sensing, environmental monitoring, and medical diagnostics. We present a multimodal formulation of hybrid learning and propose a cross attention mid fusion architecture in which a classical representation queries quantum derived feature tokens through an attention block with residual connectivity. The quantum branch is kept within practical NISQ budgets and uses up to nine qubits. We evaluate on Wine, Breast Cancer, Forest CoverType, FashionMNIST, and SteelPlatesFaults, comparing a quantum only model, a classical baseline, residual hybrid models, and the proposed mid fusion model under a consistent protocol. Pure quantum and standard hybrid designs underperform due to measurement induced information loss, while cross attention mid fusion is consistently competitive and improves performance on the more complex datasets in most cases. These findings suggest that quantum derived information becomes most valuable when integrated through principled multimodal fusion rather than used in isolation or loosely appended to classical features.
Abstract Hybrid quantum-classical neural networks (QCNNs) integrate principles from quantum computing principle and classical neural networks, offering a novel computational approach for image classification tasks. However, current QCNNs with sequential structures encounter limitations in accuracy and robustness, especially when dealing with tasks involving numerous classes. In this study, we propose a novel solution - the hybrid Parallel Quantum Classical Neural Network (PQCNN) - for image classification tasks. This architecture seamlessly integrates the parallel processing capabilities of quantum computing with the hierarchical feature extraction abilities of classical neural networks, aiming to overcome the constraints of conventional sequential structures in multi-class classification tasks. Extensive experimentation demonstrates the superiority of PQCNN over traditional concatenative structures in binary classification datasets, displaying heightened accuracy and robustness against noise. Particularly noteworthy is PQCNN's significantly improved accuracy on datasets with 5 and 10 classes. These findings underscore the transformative potential of the PQCNN architecture as an advanced solution for enhancing the performance of quantum-classical-based classifiers, particularly in the domain of image classification.
Hybrid quantum-classical neural networks provide a promising approach for incorporating quantum circuits into machine learning in the noisy intermediate-scale quantum regime. However, existing hybrid models often rely on fixed or globally shared fusion strategies, which may limit their ability to exploit complementary information carried by quantum branches, especially under distribution shifts. In this work, we propose a Feature-Adaptive Fusion Hybrid Quantum-Classical Neural Network (FAF-HQNN) for biomedical image classification. The model combines a classical deep feature encoder with a variational quantum circuit (VQC) and introduces a feature-adaptive fusion mechanism to dynamically weight classical and quantum predictions. We evaluate FAF-HQNN on two MedMNIST benchmarks, PathMNIST and BloodMNIST, under clean and corrupted test conditions. FAF-HQNN achieves the strongest overall performance on clean data among the compared methods and shows improved robustness under Gaussian, salt-and-pepper, and Poisson corruptions. Further analysis of circuit layout, measurement basis, and depth shows that even shallow variational quantum circuits can provide useful complementary information. These results demonstrate that feature-adaptive fusion is an effective strategy for improving accuracy and robustness in hybrid quantum-classical models for biomedical image classification.
The Unit Commitment Problem (UCP) is a core optimization task for ensuring secure and economically efficient operation of power grids. However, existing methodologies face two major limitations: first, there is a lack of a general and efficient way to represent the complex relationships between UCP variables and constraints; second, exponential parameter growth with problem scale impedes characterization of complex dependencies in UCP, thereby limiting predictive accuracy. To address these issues, we propose a hybrid quantum-classical graph neural network framework (HQGNN) that establishes a novel paradigm for UCP by: (1) modeling UCP as a unified bipartite graph format, and (2) integrating quantum feature extraction modules to reduce the parameter size and enhance classical GNNs’ capability in capturing high-dimensional structural characteristics. Extensive experiments on IEEE and RTS benchmark datasets demonstrate that HQGNN achieves improvements in UCP solution accuracy while validating the effectiveness of quantum modules in enhancing graph-structured feature learning. The quantum module in HQGNN is proven to be the key enabler of performance gains. Its integration cuts parameters by 16.8% and reduces mean squared error (MSE) in optimal solution prediction by 24.4% across all benchmarks, while achieving full-task superiority on the IEEE-9 dataset.These results substantiate the proposed method’s capacity to simultaneously enhance solution quality and computational efficiency through quantum-enhanced graph representation learning.