Yingxin Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yingxin Liu has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
- Ling-Li Zeng
National University of Defense Technology
5 shared publications - Dewen Hu
National university of defense technology, Changsha, China
4 shared publications Yang Yu
College of Intelligence Science and Technology, National University of Defense Technology , Changsha 410073
3 shared publications- Yang Yu
College of Intelligence Science and Technology, National University of Defense Technology, Changsha, China
3 shared publications Zongtan Zhou
College of Intelligence Science and Technology, National University of Defense Technology , Changsha 410073
3 shared publications- Zongtan Zhou
National University of Defense Technology
3 shared publications
Latest Publications
- Open Access
- 23 October 2026
- Published by Oxford University Press (OUP) in National Science Review
The Evolution of Brain-Computer Interfaces Towards Collaborative Intelligence
Brain-computer interfaces (BCIs) are evolving from unidirectional control toward deeper integration of biological and machine intelligence. This transition is constrained by two complementary bottlenecks: Intent Mismatch, or Brain-to-Machine Underreach, in which user intentions cannot be sufficiently decoded, represented, or executed by the machine; and State Mismatch, or Machine-to-Brain Overload, in which system demands exceed the user’s current cognitive, affective, or neural resources. This review uses these two mismatches as an organizing framework for examining the evolution of BCIs toward brain-machine collaborative intelligence, while distinguishing non-invasive and invasive modalities according to their signal characteristics and translational constraints. We review neural decoding, generative artificial intelligence, and shared control as approaches for reducing Intent Mismatch, and passive BCIs, adaptive interaction, neurofeedback, and neuromodulation as approaches for reducing State Mismatch. By integrating these solution paths, this review outlines a pathway toward adaptive, bidirectional, and human-centered BCI systems.
- 23 October 2026
- Published by Oxford University Press (OUP) in National Science Review
Open AccessThe Evolution of Brain-Computer Interfaces Towards Collaborative Intelligence
Brain-computer interfaces (BCIs) are evolving from unidirectional control toward deeper integration of biological and machine intelligence. This transition is constrained by two complementary bottlenecks: Intent Mismatch, or Brain-to-Machine Underreach, in which user intentions cannot be sufficiently decoded, represented, or executed by the machine; and State Mismatch, or Machine-to-Brain Overload, in which system demands exceed the user’s current cognitive, affective, or neural resources. This review uses these two mismatches as an organizing framework for examining the evolution of BCIs toward brain-machine collaborative intelligence, while distinguishing non-invasive and invasive modalities according to their signal characteristics and translational constraints. We review neural decoding, generative artificial intelligence, and shared control as approaches for reducing Intent Mismatch, and passive BCIs, adaptive interaction, neurofeedback, and neuromodulation as approaches for reducing State Mismatch. By integrating these solution paths, this review outlines a pathway toward adaptive, bidirectional, and human-centered BCI systems.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 56 (5) , 3235-3249
Decoding Driving Intentions via a Novel Brain–Computer Interface Paradigm With Low Cognitive Load and High Robustness
In recent years, brain–computer interface (BCI) based on electroencephalography (EEG) has been increasingly applied in human–vehicle collaborative driving. In this article, we design a novel BCI paradigm, incorporating subliminal steady state visual evoked potential (SSVEP) within the friendly interaction framework of short driving videos, which ensures low cognitive load interaction for drivers while also enhancing the robustness of EEG decoding. To robustly decode these signals, we propose a novel multidomain spatial–frequency–temporal multiscale gating convolutional neural network (SFT-GCNN), which explicitly addresses EEG nonstationarity and subject variability through three key innovations: 1) a channel-wise attention mechanism to extract task-relevant spatial topologies; 2) a multiscale gating convolutional unit (GCU) that adaptively filters noise and captures temporal dynamics across diverse receptive fields; and 3) a multiview fusion strategy integrating spatial, temporal, and spectral features under the joint supervision of cross-entropy (CE) and center loss to enforce intraclass compactness. The proposed decoding method outperforms several benchmark methods, achieving accuracies of 82.91% $\pm ~4.35$ % and 78.23% $\pm ~1.87$ % in the subject-dependent and subject-independent experiments. Furthermore, the subjective fatigue scales and a mean theta-to-alpha ratio (TAR) of 1.13 confirm that the proposed stimulus paradigm does not induce additional visual fatigue to participants. These results demonstrate that our approach effectively balances high decoding robustness with user comfort in practical driving scenarios.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 56 (5) , 3235-3249
Decoding Driving Intentions via a Novel Brain–Computer Interface Paradigm With Low Cognitive Load and High Robustness
In recent years, brain–computer interface (BCI) based on electroencephalography (EEG) has been increasingly applied in human–vehicle collaborative driving. In this article, we design a novel BCI paradigm, incorporating subliminal steady state visual evoked potential (SSVEP) within the friendly interaction framework of short driving videos, which ensures low cognitive load interaction for drivers while also enhancing the robustness of EEG decoding. To robustly decode these signals, we propose a novel multidomain spatial–frequency–temporal multiscale gating convolutional neural network (SFT-GCNN), which explicitly addresses EEG nonstationarity and subject variability through three key innovations: 1) a channel-wise attention mechanism to extract task-relevant spatial topologies; 2) a multiscale gating convolutional unit (GCU) that adaptively filters noise and captures temporal dynamics across diverse receptive fields; and 3) a multiview fusion strategy integrating spatial, temporal, and spectral features under the joint supervision of cross-entropy (CE) and center loss to enforce intraclass compactness. The proposed decoding method outperforms several benchmark methods, achieving accuracies of 82.91% $\pm ~4.35$ % and 78.23% $\pm ~1.87$ % in the subject-dependent and subject-independent experiments. Furthermore, the subjective fatigue scales and a mean theta-to-alpha ratio (TAR) of 1.13 confirm that the proposed stimulus paradigm does not induce additional visual fatigue to participants. These results demonstrate that our approach effectively balances high decoding robustness with user comfort in practical driving scenarios.
- 26 November 2024
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 101
Recognizing drivers’ turning intentions with EEG and eye movement
- 26 November 2024
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 101
Recognizing drivers’ turning intentions with EEG and eye movement
- Open Access
- 6 August 2024
- Published by Elsevier in Biocybernetics and Biomedical Engineering
- Vol. 44 (3) , 470-480
Parallel collaboration and closed-loop control of a cursor using multimodal physiological signals
- 6 August 2024
- Published by Elsevier in Biocybernetics and Biomedical Engineering
- Vol. 44 (3) , 470-480
Open AccessParallel collaboration and closed-loop control of a cursor using multimodal physiological signals
- 22 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (5) , 3282-3292
Cognitive Load Prediction From Multimodal Physiological Signals Using Multiview Learning
Predicting cognitive load is a crucial issue in the emerging field of human-computer interaction and holds significant practical value, particularly in flight scenarios. Although previous studies have realized efficient cognitive load classification, new research is still needed to adapt the current state-of-the-art multimodal fusion methods. Here, we proposed a feature selection framework based on multiview learning to address the challenges of information redundancy and reveal the common physiological mechanisms underlying cognitive load. Specifically, the multimodal signal features (EEG, EDA, ECG, EOG, & eye movements) at three cognitive load levels were estimated during multiattribute task battery (MATB) tasks performed by 22 healthy participants and fed into a feature selection-multiview classification with cohesion and diversity (FS-MCCD) framework. The optimized feature set was extracted from the original feature set by integrating the weight of each view and the feature weights to formulate the ranking criteria. The cognitive load prediction model, evaluated using real-time classification results, achieved an average accuracy of 81.08% and an average F1-score of 80.94% for three-class classification among 22 participants. Furthermore, the weights of the physiological signal features revealed the physiological mechanisms related to cognitive load. Specifically, heightened cognitive load was linked to amplified δ and θ power in the frontal lobe, reduced α power in the parietal lobe, and an increase in pupil diameter. Thus, the proposed multimodal feature fusion framework emphasizes the effectiveness and efficiency of using these features to predict cognitive load.
- 22 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (5) , 3282-3292
Cognitive Load Prediction From Multimodal Physiological Signals Using Multiview Learning
Predicting cognitive load is a crucial issue in the emerging field of human-computer interaction and holds significant practical value, particularly in flight scenarios. Although previous studies have realized efficient cognitive load classification, new research is still needed to adapt the current state-of-the-art multimodal fusion methods. Here, we proposed a feature selection framework based on multiview learning to address the challenges of information redundancy and reveal the common physiological mechanisms underlying cognitive load. Specifically, the multimodal signal features (EEG, EDA, ECG, EOG, & eye movements) at three cognitive load levels were estimated during multiattribute task battery (MATB) tasks performed by 22 healthy participants and fed into a feature selection-multiview classification with cohesion and diversity (FS-MCCD) framework. The optimized feature set was extracted from the original feature set by integrating the weight of each view and the feature weights to formulate the ranking criteria. The cognitive load prediction model, evaluated using real-time classification results, achieved an average accuracy of 81.08% and an average F1-score of 80.94% for three-class classification among 22 participants. Furthermore, the weights of the physiological signal features revealed the physiological mechanisms related to cognitive load. Specifically, heightened cognitive load was linked to amplified δ and θ power in the frontal lobe, reduced α power in the parietal lobe, and an increase in pupil diameter. Thus, the proposed multimodal feature fusion framework emphasizes the effectiveness and efficiency of using these features to predict cognitive load.
- 19 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 53 (2) , 357-366
Fusion of Spatial, Temporal, and Spectral EEG Signatures Improves Multilevel Cognitive Load Prediction
Cognitive load prediction is one of the most important issues in the nascent field of neuroergonomics, and it has significant value in real-world applications. Most of the previous studies of cognitive load prediction only utilized electroencephalography (EEG)-based spectral signatures or interchannel connectivity, ignoring abundant temporal microstate features, which may represent the transient topologies of EEG signals. Furthermore, previous studies have mostly focused on the binary-level classification of cognitive load for single-type cognitive tasks. To date, there are few studies on the multilevel prediction of cognitive load during mixed cognitive tasks. Here, we first designed a new paradigm termed the “finding fault game,” mixing multiple tasks of memory, counting, and visual search, and then developed a multidimensional analysis framework to improve cognitive load prediction using a fusion of spatial, temporal, and spectral EEG features. Specifically, EEG-based functional connectivity, microstates and power spectral densities (PSD) were calculated for three cognitive load levels. Twelve adult subjects participated in the study. The experimental results show that increased cognitive load was associated with elevated theta and degraded alpha power and significant changes in interchannel connectivity and microstates, and that fusing the three types of EEG features improved the performance of three-level cognitive load prediction, achieving the accuracies of greater than 80% in the cross-validation, real-time, and over-time prediction. The findings suggest that all three types of EEG features can serve as signatures of cognitive load and that their fusion can improve multilevel prediction.
- 19 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 53 (2) , 357-366
Fusion of Spatial, Temporal, and Spectral EEG Signatures Improves Multilevel Cognitive Load Prediction
Cognitive load prediction is one of the most important issues in the nascent field of neuroergonomics, and it has significant value in real-world applications. Most of the previous studies of cognitive load prediction only utilized electroencephalography (EEG)-based spectral signatures or interchannel connectivity, ignoring abundant temporal microstate features, which may represent the transient topologies of EEG signals. Furthermore, previous studies have mostly focused on the binary-level classification of cognitive load for single-type cognitive tasks. To date, there are few studies on the multilevel prediction of cognitive load during mixed cognitive tasks. Here, we first designed a new paradigm termed the “finding fault game,” mixing multiple tasks of memory, counting, and visual search, and then developed a multidimensional analysis framework to improve cognitive load prediction using a fusion of spatial, temporal, and spectral EEG features. Specifically, EEG-based functional connectivity, microstates and power spectral densities (PSD) were calculated for three cognitive load levels. Twelve adult subjects participated in the study. The experimental results show that increased cognitive load was associated with elevated theta and degraded alpha power and significant changes in interchannel connectivity and microstates, and that fusing the three types of EEG features improved the performance of three-level cognitive load prediction, achieving the accuracies of greater than 80% in the cross-validation, real-time, and over-time prediction. The findings suggest that all three types of EEG features can serve as signatures of cognitive load and that their fusion can improve multilevel prediction.
- 12 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 15 (4) , 1722-1731
Graph Learning With Co-Teaching for EEG-Based Motor Imagery Recognition
Previous studies have explored the use of deep neural networks for electroencephalography-based motor imagery recognition, but most of the models focus on the recognition performance achieved for a single subject and are challenging to transfer due to individual differences and low signal-to-noise-ratio of EEG signals. To date, few studies have paid attention to the balance between generalizability and personalization across subjects. To this end, we propose a co-teaching graph learning method for cross-subject EEG-based motor imagery recognition. First, A novel graph learning approach is designed to improve feature extraction from a typical graph structure containing raw EEG signals. Second, two graph learning models are constructed to filter noisy data by using a co-teaching training strategy, preventing overfitting on noisy samples obtained from different subjects. The proposed model shows a 5.4% and 3.2% increase in accuracy of single-and multi-subject four-class motor imagery recognition tasks compared to the previous best method, respectively. Experimental results also demonstrate that it is easy to derive a model that can represent generic knowledge of multiple motor imagery subjects and can be fine-tuned efficiently for new subjects.
- 12 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cognitive and Developmental Systems
- Vol. 15 (4) , 1722-1731
Graph Learning With Co-Teaching for EEG-Based Motor Imagery Recognition
Previous studies have explored the use of deep neural networks for electroencephalography-based motor imagery recognition, but most of the models focus on the recognition performance achieved for a single subject and are challenging to transfer due to individual differences and low signal-to-noise-ratio of EEG signals. To date, few studies have paid attention to the balance between generalizability and personalization across subjects. To this end, we propose a co-teaching graph learning method for cross-subject EEG-based motor imagery recognition. First, A novel graph learning approach is designed to improve feature extraction from a typical graph structure containing raw EEG signals. Second, two graph learning models are constructed to filter noisy data by using a co-teaching training strategy, preventing overfitting on noisy samples obtained from different subjects. The proposed model shows a 5.4% and 3.2% increase in accuracy of single-and multi-subject four-class motor imagery recognition tasks compared to the previous best method, respectively. Experimental results also demonstrate that it is easy to derive a model that can represent generic knowledge of multiple motor imagery subjects and can be fine-tuned efficiently for new subjects.