Bin Lu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Bin Lu has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Fuwang Wang
School of Mechanical Engineering, Northeast Electric Power University, Jilin, China
5 shared publications - Rongrong Fu
Hebei Key Laboratory of Measurement Technology and Instrumentation, Yanshan University, Qinhuangdao, China
5 shared publications - Changchun Hua
Yanshan University
3 shared publications - Guilin Wen
School of Mechanical Engineering, Yanshan University, Qinhuangdao, China
2 shared publications - Junxiang Chen
Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, PA, USA
2 shared publications - Daping Chen
School of Mechanic Engineering, Northeast Electric Power University, Jilin, China
1 shared publication
Latest Publications
- 30 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. PP (15513203) , 1-11
Motor Intention Recognition via Dual-Supervised Disentangled Representation Learning With Cross-Latent Swapping Alignment
Motor imagery-based brain-computer interfaces (BCIs) hold substantial promise for neurorehabilitation and assistive control, yet their practical deployment is constrained by pronounced intersubject variability of electroencephalographic (EEG) signals. To address this challenge, this article proposed a cross-aligned disentangled latent framework (CAD-LF) for learning identifiable and semantically disentangled representations. The framework first employed a dual-guided disentangled identifiable variational autoencoder to explicitly disentangle task-relevant and subject-specific identifiable latent factors under auxiliary supervision. To enable inference without auxiliary variables, a latent-aligned transfer autoencoder was introduced to inherit disentangled semantics via cross-module latent alignment within a cyclic encoder-decoder architecture. Additionally, a disentangled factor permutation mechanism further enforced structural consistency and improved disentanglement robustness. Experiments on BCI Competition IV Dataset 2a demonstrated that CAD-LF achieved 79.02% accuracy (k = 0.72) in cross-session evaluation and 66.06% (k = 0.55) in the more challenging cross-subject setting, outperforming competitive baselines. These results indicate that the proposed framework effectively mitigates confounding interference, enhances representation stability, and improves out-of-distribution generalization, offering a principled approach for robust EEG decoding.
- 30 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. PP (15513203) , 1-11
Motor Intention Recognition via Dual-Supervised Disentangled Representation Learning With Cross-Latent Swapping Alignment
Motor imagery-based brain-computer interfaces (BCIs) hold substantial promise for neurorehabilitation and assistive control, yet their practical deployment is constrained by pronounced intersubject variability of electroencephalographic (EEG) signals. To address this challenge, this article proposed a cross-aligned disentangled latent framework (CAD-LF) for learning identifiable and semantically disentangled representations. The framework first employed a dual-guided disentangled identifiable variational autoencoder to explicitly disentangle task-relevant and subject-specific identifiable latent factors under auxiliary supervision. To enable inference without auxiliary variables, a latent-aligned transfer autoencoder was introduced to inherit disentangled semantics via cross-module latent alignment within a cyclic encoder-decoder architecture. Additionally, a disentangled factor permutation mechanism further enforced structural consistency and improved disentanglement robustness. Experiments on BCI Competition IV Dataset 2a demonstrated that CAD-LF achieved 79.02% accuracy (k = 0.72) in cross-session evaluation and 66.06% (k = 0.55) in the more challenging cross-subject setting, outperforming competitive baselines. These results indicate that the proposed framework effectively mitigates confounding interference, enhances representation stability, and improves out-of-distribution generalization, offering a principled approach for robust EEG decoding.
- 27 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (3) , 2253-2264
Causality-Driven Convolutional Manifold Attention Network for Electroencephalogram Signal Decoding
Deep learning-based methods have achieved remarkable success in brain-computer interfaces (BCIs). However, its inherent assumption of independent and identically distributed (i.i.d.) data renders it vulnerable to out-of-distribution (OOD) scenarios. To address this limitation, the present study proposed a causality-driven convolutional manifold attention network (CD-CMAN) that learned invariant representations from electroencephalogram (EEG) signals to enhance OOD generalization. The framework began with a spatiotemporal convolution module to extract rich temporal and spatial features. Guided by the defined structural causal model and leveraging the strengths of Riemannian geometry and deep learning, dual latent encoders with manifold attention units were crafted to explicitly separate spatiotemporal feature maps into semantic and variation latent factors. A reconstruction module with a dedicated loss was implemented to ensure these factors retaining informative, while the Hilbert-Schmidt independence criterion (HSIC) was introduced to enforce their statistical independence. Further, a variational information bottleneck and gradient reversal layer were incorporated to compress and disentangle the semantic and variation factors. Evaluations on two public datasets under both subject-dependent and subject-independent settings demonstrated that CD-CMAN consistently outperforms comparative baselines. These findings suggest that the proposed model could provide a new solution for the practical application of BCI technology.
- 27 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 48 (3) , 2253-2264
Causality-Driven Convolutional Manifold Attention Network for Electroencephalogram Signal Decoding
Deep learning-based methods have achieved remarkable success in brain-computer interfaces (BCIs). However, its inherent assumption of independent and identically distributed (i.i.d.) data renders it vulnerable to out-of-distribution (OOD) scenarios. To address this limitation, the present study proposed a causality-driven convolutional manifold attention network (CD-CMAN) that learned invariant representations from electroencephalogram (EEG) signals to enhance OOD generalization. The framework began with a spatiotemporal convolution module to extract rich temporal and spatial features. Guided by the defined structural causal model and leveraging the strengths of Riemannian geometry and deep learning, dual latent encoders with manifold attention units were crafted to explicitly separate spatiotemporal feature maps into semantic and variation latent factors. A reconstruction module with a dedicated loss was implemented to ensure these factors retaining informative, while the Hilbert-Schmidt independence criterion (HSIC) was introduced to enforce their statistical independence. Further, a variational information bottleneck and gradient reversal layer were incorporated to compress and disentangle the semantic and variation factors. Evaluations on two public datasets under both subject-dependent and subject-independent settings demonstrated that CD-CMAN consistently outperforms comparative baselines. These findings suggest that the proposed model could provide a new solution for the practical application of BCI technology.
- 25 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 55 (5) , 2202-2212
Dynamic Hierarchical Convolutional Attention Network for Recognizing Motor Imagery Intention
The neural activity patterns of localized brain regions are crucial for recognizing brain intentions. However, existing electroencephalogram (EEG) decoding models, especially those based on deep learning, predominantly focus on global spatial features, neglecting valuable local information, potentially leading to suboptimal performance. Therefore, this study proposed a dynamic hierarchical convolutional attention network (DH-CAN) that comprehensively learned discriminative information from both global and local spatial domains, as well as from time-frequency domains in EEG signals. Specifically, a multiscale convolutional block was designed to dynamically capture time-frequency information. The channels of EEG signals were mapped to different brain regions based on motor imagery neural activity patterns. The spatial features, both global and local, were then hierarchically extracted to fully exploit the discriminative information. Furthermore, regional connectivity was established using a graph attention network, incorporating it into the local spatial features. Particularly, this study shared network parameters between symmetrical brain regions to better capture asymmetrical motor imagery patterns. Finally, the learned multilevel features were integrated through a high-level fusion layer. Extensive experimental results on two datasets demonstrated that the proposed model performed excellently across multiple evaluation metrics, exceeding existing benchmark methods. These findings suggested that the proposed model offered a novel perspective for EEG decoding research.
- 25 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 55 (5) , 2202-2212
Dynamic Hierarchical Convolutional Attention Network for Recognizing Motor Imagery Intention
The neural activity patterns of localized brain regions are crucial for recognizing brain intentions. However, existing electroencephalogram (EEG) decoding models, especially those based on deep learning, predominantly focus on global spatial features, neglecting valuable local information, potentially leading to suboptimal performance. Therefore, this study proposed a dynamic hierarchical convolutional attention network (DH-CAN) that comprehensively learned discriminative information from both global and local spatial domains, as well as from time-frequency domains in EEG signals. Specifically, a multiscale convolutional block was designed to dynamically capture time-frequency information. The channels of EEG signals were mapped to different brain regions based on motor imagery neural activity patterns. The spatial features, both global and local, were then hierarchically extracted to fully exploit the discriminative information. Furthermore, regional connectivity was established using a graph attention network, incorporating it into the local spatial features. Particularly, this study shared network parameters between symmetrical brain regions to better capture asymmetrical motor imagery patterns. Finally, the learned multilevel features were integrated through a high-level fusion layer. Extensive experimental results on two datasets demonstrated that the proposed model performed excellently across multiple evaluation metrics, exceeding existing benchmark methods. These findings suggested that the proposed model offered a novel perspective for EEG decoding research.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-16
ECG-Based Real-Time Drivers’ Fatigue Detection Using a Novel Elastic Dry Electrode
In view of the complicated placement process of wet electrodes used for driving fatigue detection based on electrocardiogram (ECG) characteristics, this study proposed a novel elastic dry electrode for real-time acquisition of ECG signals from the palm of hand. This type of dry electrode eliminated the need for applying conductive paste when collecting the driver’s ECG. Meanwhile, it avoids the requirement of penetrating the cuticle layer of the skin surface to reduce impedance as well, thus minimizing the risk of skin infection. To alleviate driver fatigue resulting from prolonged driving, a periodic man–machine interaction mode (PMIM) was proposed. To assess the efficacy of PMIM in mitigating driver fatigue, this study compared and evaluated driver mental fatigue based on electroencephalogram (EEG) and heart rate variability (HRV) characteristics across two driving modes. The findings demonstrated that the novel elastic dry electrode effectively captured changes in driving fatigue characteristics by acquiring ECG signals from the palm of the hand. In addition, the proposed PMIM effectively mitigated driving fatigue resulting from prolonged periods of driving.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-16
ECG-Based Real-Time Drivers’ Fatigue Detection Using a Novel Elastic Dry Electrode
In view of the complicated placement process of wet electrodes used for driving fatigue detection based on electrocardiogram (ECG) characteristics, this study proposed a novel elastic dry electrode for real-time acquisition of ECG signals from the palm of hand. This type of dry electrode eliminated the need for applying conductive paste when collecting the driver’s ECG. Meanwhile, it avoids the requirement of penetrating the cuticle layer of the skin surface to reduce impedance as well, thus minimizing the risk of skin infection. To alleviate driver fatigue resulting from prolonged driving, a periodic man–machine interaction mode (PMIM) was proposed. To assess the efficacy of PMIM in mitigating driver fatigue, this study compared and evaluated driver mental fatigue based on electroencephalogram (EEG) and heart rate variability (HRV) characteristics across two driving modes. The findings demonstrated that the novel elastic dry electrode effectively captured changes in driving fatigue characteristics by acquiring ECG signals from the palm of the hand. In addition, the proposed PMIM effectively mitigated driving fatigue resulting from prolonged periods of driving.
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (16) , 17891-17900
A Novel Semi-Dry Electrode Suitable for Collecting EEG Data for Detecting Driving Fatigue in Long-Period Driving Case
To address issues with the conventional wet electrode during long-period experiments, which include signal acquisition distortion caused by conductive liquid volatilization and drying, as well as time-consuming conductive liquid refilling, this study introduced a portable semi-dry electrode that allows for convenient conductive liquid replenishment at any time. The new semi-dry electrode combines the advantages of small impedance value of wet electrode and convenient installation and disassembly of dry electrode. And this electrode is suitable for collecting human electroencephalogram (EEG) for a long time, such as detecting the EEG characteristics of drivers who drive vehicles for a long time. In addition, since the order recurrence plot (ORP) does not require for signal stability, data length, and signal-to-noise ratio, this paper used the method to analyze the driving fatigue characteristics of the subjects. Firstly, the ORP method was used to map EEG signals collected by the two types of electrodes, and the ORP was performed by recurrence quantitative analysis (RQA). Then, two quantitative parameters, determinism (DET) and average diagonal line length (DLL), were selected to analyze the fatigue driving characteristics of the subjects. The results showed that with the increase of fatigue degree of the subjects, the blue and white squares in ORP from EEG signals collected by the two kinds of electrodes gradually increased, with a clearer texture and a more organized arrangement. Meanwhile, the quantization parameters DET and DLL increase gradually as well, which can effectively express the law of driving fatigue change. The results of this study are helpful to address the problems such as the difficulty of supplementing conductive fluid with traditional wet electrodes, and to address the problem of easy interference in EEG acquisition experiments in complex environments.
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 23 (16) , 17891-17900
A Novel Semi-Dry Electrode Suitable for Collecting EEG Data for Detecting Driving Fatigue in Long-Period Driving Case
To address issues with the conventional wet electrode during long-period experiments, which include signal acquisition distortion caused by conductive liquid volatilization and drying, as well as time-consuming conductive liquid refilling, this study introduced a portable semi-dry electrode that allows for convenient conductive liquid replenishment at any time. The new semi-dry electrode combines the advantages of small impedance value of wet electrode and convenient installation and disassembly of dry electrode. And this electrode is suitable for collecting human electroencephalogram (EEG) for a long time, such as detecting the EEG characteristics of drivers who drive vehicles for a long time. In addition, since the order recurrence plot (ORP) does not require for signal stability, data length, and signal-to-noise ratio, this paper used the method to analyze the driving fatigue characteristics of the subjects. Firstly, the ORP method was used to map EEG signals collected by the two types of electrodes, and the ORP was performed by recurrence quantitative analysis (RQA). Then, two quantitative parameters, determinism (DET) and average diagonal line length (DLL), were selected to analyze the fatigue driving characteristics of the subjects. The results showed that with the increase of fatigue degree of the subjects, the blue and white squares in ORP from EEG signals collected by the two kinds of electrodes gradually increased, with a clearer texture and a more organized arrangement. Meanwhile, the quantization parameters DET and DLL increase gradually as well, which can effectively express the law of driving fatigue change. The results of this study are helpful to address the problems such as the difficulty of supplementing conductive fluid with traditional wet electrodes, and to address the problem of easy interference in EEG acquisition experiments in complex environments.