Yaochong Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yaochong Li has a total of 53 co-authors. Here are the top co-authors by number of shared publications:
- Ri-Gui Zhou
Shanghai Maritime University
22 shared publications - Ruiqing Xu
Faculty of Intelligence Technology, Shanghai Institute of Technology, Shanghai, People’s Republic of China
7 shared publications WenWen Hu
4 shared publicationsJia Luo
College of Information Engineering, Shanghai Maritime University, Shanghai, China
2 shared publications- Jia Luo
Shangrao Normal College
2 shared publications - Pengju Ren
College of Information Engineering, Shanghai Maritime University, Shanghai, China
2 shared publications
Latest Publications
- 15 July 2026
- Published by Elsevier in Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
- Vol. 363 (Pt 1) , 128416
MSAMixerNet: A multi-scale MLP-Mixer model with attention mechanisms for Raman spectral classification of animal leather
- 15 July 2026
- Published by Elsevier in Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
- Vol. 363 (Pt 1) , 128416
MSAMixerNet: A multi-scale MLP-Mixer model with attention mechanisms for Raman spectral classification of animal leather
- 5 March 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (9) , 095104
Measurement-device-independent semi-quantum dialogue protocol based on hyper-entangled Bell states
To address the dual challenges of scarce quantum resources and measurement device vulnerabilities in practical quantum dialogue (QD) protocols, this paper proposes a measurement-device-independent semi-quantum dialogue (MDI-SQD) protocol based on hyperentangled Bell states. Employing a dual-degree-of-freedom encoding strategy, the protocol enables both parties to simultaneously encode secret messages and authentication information across their respective degrees of freedom, achieving identity verification and message transmission in a single round. To counter side-channel attacks, the protocol further incorporates a MDI framework, delegating critical measurement steps to untrusted third parties to effectively safeguard protocol security. Theoretical analysis demonstrates robust resistance to attacks. Performance evaluations reveal communication efficiency reaching 75%, significantly outperforming existing semi-quantum dialogue protocols.
- 5 March 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (9) , 095104
Measurement-device-independent semi-quantum dialogue protocol based on hyper-entangled Bell states
To address the dual challenges of scarce quantum resources and measurement device vulnerabilities in practical quantum dialogue (QD) protocols, this paper proposes a measurement-device-independent semi-quantum dialogue (MDI-SQD) protocol based on hyperentangled Bell states. Employing a dual-degree-of-freedom encoding strategy, the protocol enables both parties to simultaneously encode secret messages and authentication information across their respective degrees of freedom, achieving identity verification and message transmission in a single round. To counter side-channel attacks, the protocol further incorporates a MDI framework, delegating critical measurement steps to untrusted third parties to effectively safeguard protocol security. Theoretical analysis demonstrates robust resistance to attacks. Performance evaluations reveal communication efficiency reaching 75%, significantly outperforming existing semi-quantum dialogue protocols.
- 23 February 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (8) , 086002
Point cloud symmetry encoding and multi-channel rotational equivariant quantum neural networks
Equivariant Quantum Neural Networks can mitigate the barren plateau problem by embedding inductive bias into variational quantum circuits, making them well-suited for processing symmetric point clouds. However, existing point cloud to quantum state encoding methods fail to preserve inherent symmetries of point clouds, and research on compatible equivariant quantum circuits remains inadequate. To address these limitations, this paper proposes cylindrical and spherical voxelization encoding for point clouds, which retains rotational and vertical cyclic symmetry about the Z-axis. Concurrently, Multi-Channel Rotational Equivariant Quantum Neural Networks are designed, where symmetries across different dimensions are treated as distinct channels. Experimental results demonstrate that cylindrical voxelization achieves superior performance in encoding point clouds. Specifically, this work attains a comparable level of accuracy with a smaller parameter scale than Hardware Efficient Ansatz and Rotational Equivariant Quantum Neural Networks, while also alleviating the barren plateau problem. These findings demonstrate the applicability of proposed encoding method, reveal the adaptive relationship between encoding characteristics and model architecture, and provide a potentially feasible solution for processing complex point clouds with large-scale qubits.
- 23 February 2026
- Published by IOP Publishing in Physica Scripta
- Vol. 101 (8) , 086002
Point cloud symmetry encoding and multi-channel rotational equivariant quantum neural networks
Equivariant Quantum Neural Networks can mitigate the barren plateau problem by embedding inductive bias into variational quantum circuits, making them well-suited for processing symmetric point clouds. However, existing point cloud to quantum state encoding methods fail to preserve inherent symmetries of point clouds, and research on compatible equivariant quantum circuits remains inadequate. To address these limitations, this paper proposes cylindrical and spherical voxelization encoding for point clouds, which retains rotational and vertical cyclic symmetry about the Z-axis. Concurrently, Multi-Channel Rotational Equivariant Quantum Neural Networks are designed, where symmetries across different dimensions are treated as distinct channels. Experimental results demonstrate that cylindrical voxelization achieves superior performance in encoding point clouds. Specifically, this work attains a comparable level of accuracy with a smaller parameter scale than Hardware Efficient Ansatz and Rotational Equivariant Quantum Neural Networks, while also alleviating the barren plateau problem. These findings demonstrate the applicability of proposed encoding method, reveal the adaptive relationship between encoding characteristics and model architecture, and provide a potentially feasible solution for processing complex point clouds with large-scale qubits.
- Open Access
- 1 January 2026
- Published by IOP Publishing in New Journal of Physics
- Vol. 28 (1) , 014511
Dual-regularized nonlinear quantum encoding for adversarial robustness in quantum machine learning
Quantum machine learning (QML) models are highly vulnerable to adversarial attacks, severely limiting their deployment in security-sensitive applications. To enhance their robustness, this study proposes a novel method, termed nonlinear quantum encoding with dual regularisation (NQE-DR), which combines nonlinear quantum encoding with dual regularisation. NQE-DR integrates classical nonlinear transformations, adaptive parameter optimisation, and a multi-qubit cross-coupling architecture. Our dual regularisation incorporates parameter and gradient regularisation, providing a theoretical framework to guarantee robustness under adversarial attacks and noisy conditions. Experimental results demonstrate that NQE-DR significantly outperforms mainstream encoding methods, exhibiting a notable improvement in classification accuracy and enhanced stability under FGSM attacks and noise. This study advances adversarial quantum machine learning by seamlessly integrating nonlinear encoding with a comprehensive dual regularisation strategy, offering a potent security enhancement for practical QML deployment.
- 1 January 2026
- Published by IOP Publishing in New Journal of Physics
- Vol. 28 (1) , 014511
Open AccessDual-regularized nonlinear quantum encoding for adversarial robustness in quantum machine learning
Quantum machine learning (QML) models are highly vulnerable to adversarial attacks, severely limiting their deployment in security-sensitive applications. To enhance their robustness, this study proposes a novel method, termed nonlinear quantum encoding with dual regularisation (NQE-DR), which combines nonlinear quantum encoding with dual regularisation. NQE-DR integrates classical nonlinear transformations, adaptive parameter optimisation, and a multi-qubit cross-coupling architecture. Our dual regularisation incorporates parameter and gradient regularisation, providing a theoretical framework to guarantee robustness under adversarial attacks and noisy conditions. Experimental results demonstrate that NQE-DR significantly outperforms mainstream encoding methods, exhibiting a notable improvement in classification accuracy and enhanced stability under FGSM attacks and noise. This study advances adversarial quantum machine learning by seamlessly integrating nonlinear encoding with a comprehensive dual regularisation strategy, offering a potent security enhancement for practical QML deployment.
- 1 November 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (11) , 115101
QPCNet: A hybrid quantum positional encoding and channel attention network for image classification
Quantum-classical hybrid neural networks have recently emerged as a prominent research direction, combining the expressive power of quantum computing with the robustness and scalability of classical deep learning. These models have demonstrated significant potential, especially in image classification scenarios. However, existing hybrid architectures often fail to jointly model spatial and channel-wise features, which may constrain their representational capacity. In this work, we propose a novel quantum-enhanced convolutional neural network architecture that integrates Quantum Positional Encoding (QPE) and Quantum Channel Attention (QCA) mechanisms. These components respectively enhance spatial and channel-wise feature representations, while ensuring compatibility with conventional training pipelines. Experiments on the MNIST and Fashion-MNIST datasets demonstrate that the proposed model consistently outperforms classical CNN baselines in classification accuracy. Furthermore, experiments on the challenging CIFAR-10 dataset further validate the competitive performance of the proposed model, highlighting its strong generalization capability. These findings underscore the practical value of integrating quantum mechanisms into deep learning and emphasize their potential to advance intelligent visual recognition systems.
- 1 November 2025
- Published by IOP Publishing in Physica Scripta
- Vol. 100 (11) , 115101
QPCNet: A hybrid quantum positional encoding and channel attention network for image classification
Quantum-classical hybrid neural networks have recently emerged as a prominent research direction, combining the expressive power of quantum computing with the robustness and scalability of classical deep learning. These models have demonstrated significant potential, especially in image classification scenarios. However, existing hybrid architectures often fail to jointly model spatial and channel-wise features, which may constrain their representational capacity. In this work, we propose a novel quantum-enhanced convolutional neural network architecture that integrates Quantum Positional Encoding (QPE) and Quantum Channel Attention (QCA) mechanisms. These components respectively enhance spatial and channel-wise feature representations, while ensuring compatibility with conventional training pipelines. Experiments on the MNIST and Fashion-MNIST datasets demonstrate that the proposed model consistently outperforms classical CNN baselines in classification accuracy. Furthermore, experiments on the challenging CIFAR-10 dataset further validate the competitive performance of the proposed model, highlighting its strong generalization capability. These findings underscore the practical value of integrating quantum mechanisms into deep learning and emphasize their potential to advance intelligent visual recognition systems.
- Open Access
HQRNN-FD: A Hybrid Quantum Recurrent Neural Network for Fraud Detection
Detecting financial fraud is a critical aspect of modern intelligent financial systems. Despite the advances brought by deep learning in predictive accuracy, challenges persist—particularly in capturing complex, high-dimensional nonlinear features. This study introduces a novel hybrid quantum recurrent neural network for fraud detection (HQRNN-FD). The model utilizes variational quantum circuits (VQCs) incorporating angle encoding, data reuploading, and hierarchical entanglement to project transaction features into quantum state spaces, thereby facilitating quantum-enhanced feature extraction. For sequential analysis, the model integrates a recurrent neural network (RNN) with a self-attention mechanism to effectively capture temporal dependencies and uncover latent fraudulent patterns. To mitigate class imbalance, the synthetic minority over-sampling technique (SMOTE) is employed during preprocessing, enhancing both class representation and model generalizability. Experimental evaluations reveal that HQRNN-FD attains an accuracy of 0.972 on publicly available fraud detection datasets, outperforming conventional models by 2.4%. In addition, the framework exhibits robustness against quantum noise and improved predictive performance with increasing qubit numbers, validating its efficacy and scalability for imbalanced financial classification tasks.
Open AccessHQRNN-FD: A Hybrid Quantum Recurrent Neural Network for Fraud Detection
Detecting financial fraud is a critical aspect of modern intelligent financial systems. Despite the advances brought by deep learning in predictive accuracy, challenges persist—particularly in capturing complex, high-dimensional nonlinear features. This study introduces a novel hybrid quantum recurrent neural network for fraud detection (HQRNN-FD). The model utilizes variational quantum circuits (VQCs) incorporating angle encoding, data reuploading, and hierarchical entanglement to project transaction features into quantum state spaces, thereby facilitating quantum-enhanced feature extraction. For sequential analysis, the model integrates a recurrent neural network (RNN) with a self-attention mechanism to effectively capture temporal dependencies and uncover latent fraudulent patterns. To mitigate class imbalance, the synthetic minority over-sampling technique (SMOTE) is employed during preprocessing, enhancing both class representation and model generalizability. Experimental evaluations reveal that HQRNN-FD attains an accuracy of 0.972 on publicly available fraud detection datasets, outperforming conventional models by 2.4%. In addition, the framework exhibits robustness against quantum noise and improved predictive performance with increasing qubit numbers, validating its efficacy and scalability for imbalanced financial classification tasks.
- Open Access
AQEA-QAS: An Adaptive Quantum Evolutionary Algorithm for Quantum Architecture Search
Quantum neural networks (QNNs) represent an emerging technology that uses a quantum computer for neural network computations. The QNNs have demonstrated potential advantages over classical neural networks in certain tasks. As a core component of a QNN, the parameterized quantum circuit (PQC) plays a crucial role in determining the QNN’s overall performance. However, quantum circuit architectures designed manually based on experience or using specific hardware structures can suffer from inefficiency due to the introduction of redundant quantum gates, which amplifies the impact of noise on system performance. Recent studies have suggested that the advantages of quantum evolutionary algorithms (QEAs) in terms of precision and convergence speed can provide an effective solution to quantum circuit architecture-related problems. Currently, most QEAs adopt a fixed rotation mode in the evolution process, and a lack of an adaptive updating mode can cause the QEAs to fall into a local optimum and make it difficult for them to converge. To address these problems, this study proposes an adaptive quantum evolution algorithm (AQEA). First, an adaptive mechanism is introduced to the evolution process, and the strategy of combining two dynamic rotation angles is adopted. Second, to prevent the fluctuations of the population’s offspring, the elite retention of the parents is used to ensure the inheritance of good genes. Finally, when the population falls into a local optimum, a quantum catastrophe mechanism is employed to break the current population state. The experimental results show that compared with the QNN structure based on manual design and QEA search, the proposed AQEA can reduce the number of network parameters by up to 20% and increase the accuracy by 7.21%. Moreover, in noisy environments, the AQEA-optimized circuit outperforms traditional circuits in maintaining high fidelity, and its excellent noise resistance provides strong support for the reliability of quantum computing.
Open AccessAQEA-QAS: An Adaptive Quantum Evolutionary Algorithm for Quantum Architecture Search
Quantum neural networks (QNNs) represent an emerging technology that uses a quantum computer for neural network computations. The QNNs have demonstrated potential advantages over classical neural networks in certain tasks. As a core component of a QNN, the parameterized quantum circuit (PQC) plays a crucial role in determining the QNN’s overall performance. However, quantum circuit architectures designed manually based on experience or using specific hardware structures can suffer from inefficiency due to the introduction of redundant quantum gates, which amplifies the impact of noise on system performance. Recent studies have suggested that the advantages of quantum evolutionary algorithms (QEAs) in terms of precision and convergence speed can provide an effective solution to quantum circuit architecture-related problems. Currently, most QEAs adopt a fixed rotation mode in the evolution process, and a lack of an adaptive updating mode can cause the QEAs to fall into a local optimum and make it difficult for them to converge. To address these problems, this study proposes an adaptive quantum evolution algorithm (AQEA). First, an adaptive mechanism is introduced to the evolution process, and the strategy of combining two dynamic rotation angles is adopted. Second, to prevent the fluctuations of the population’s offspring, the elite retention of the parents is used to ensure the inheritance of good genes. Finally, when the population falls into a local optimum, a quantum catastrophe mechanism is employed to break the current population state. The experimental results show that compared with the QNN structure based on manual design and QEA search, the proposed AQEA can reduce the number of network parameters by up to 20% and increase the accuracy by 7.21%. Moreover, in noisy environments, the AQEA-optimized circuit outperforms traditional circuits in maintaining high fidelity, and its excellent noise resistance provides strong support for the reliability of quantum computing.
- 17 June 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 292
A self-supervised learning method for Raman spectroscopy based on masked autoencoders
- 17 June 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 292
A self-supervised learning method for Raman spectroscopy based on masked autoencoders
- Open Access
A Self-supervised Learning Method for Raman Spectroscopy based on Masked Autoencoders
Raman spectroscopy serves as a powerful and reliable tool for analyzing the chemical information of substances. The integration of Raman spectroscopy with deep learning methods enables rapid qualitative and quantitative analysis of materials. Most existing approaches adopt supervised learning methods. Although supervised learning has achieved satisfactory accuracy in spectral analysis, it is still constrained by costly and limited well-annotated spectral datasets for training. When spectral annotation is challenging or the amount of annotated data is insufficient, the performance of supervised learning in spectral material identification declines. In order to address the challenge of feature extraction from unannotated spectra, we propose a self-supervised learning paradigm for Raman Spectroscopy based on a Masked AutoEncoder, termed SMAE. SMAE does not require any spectral annotations during pre-training. By randomly masking and then reconstructing the spectral information, the model learns essential spectral features. The reconstructed spectra exhibit certain denoising properties, improving the signal-to-noise ratio (SNR) by more than twofold. Utilizing the network weights obtained from masked pre-training, SMAE achieves clustering accuracy of over 80% for 30 classes of isolated bacteria in a pathogenic bacterial dataset, demonstrating significant improvements compared to classical unsupervised methods and other state-of-the-art deep clustering methods. After fine-tuning the network with a limited amount of annotated data, SMAE achieves an identification accuracy of 83.90% on the test set, presenting competitive performance against the supervised ResNet (83.40%).
Open AccessA Self-supervised Learning Method for Raman Spectroscopy based on Masked Autoencoders
Raman spectroscopy serves as a powerful and reliable tool for analyzing the chemical information of substances. The integration of Raman spectroscopy with deep learning methods enables rapid qualitative and quantitative analysis of materials. Most existing approaches adopt supervised learning methods. Although supervised learning has achieved satisfactory accuracy in spectral analysis, it is still constrained by costly and limited well-annotated spectral datasets for training. When spectral annotation is challenging or the amount of annotated data is insufficient, the performance of supervised learning in spectral material identification declines. In order to address the challenge of feature extraction from unannotated spectra, we propose a self-supervised learning paradigm for Raman Spectroscopy based on a Masked AutoEncoder, termed SMAE. SMAE does not require any spectral annotations during pre-training. By randomly masking and then reconstructing the spectral information, the model learns essential spectral features. The reconstructed spectra exhibit certain denoising properties, improving the signal-to-noise ratio (SNR) by more than twofold. Utilizing the network weights obtained from masked pre-training, SMAE achieves clustering accuracy of over 80% for 30 classes of isolated bacteria in a pathogenic bacterial dataset, demonstrating significant improvements compared to classical unsupervised methods and other state-of-the-art deep clustering methods. After fine-tuning the network with a limited amount of annotated data, SMAE achieves an identification accuracy of 83.90% on the test set, presenting competitive performance against the supervised ResNet (83.40%).