Distinguishable spatial-spectral feature learning neural network framework for motor imagery-based brain–computer interface

Chang, Liu; Jin, Jing; Xu, Ruidong; Li, Shurui; Zuo, Cili; Sun, Hao; Wang, Xingyu; Cichocki, Andrzej

doi:10.1088/1741-2552/ac1d36

Cited by 9 publications

(6 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This indicates that FB-CGANet can yield superior and more stable performance than compared methods. Also, compared with recent methods, the cross-validation result of FB-CGANet is higher than that of the Enhanced Multimodal Fusion method with mean accuracy of 83.15% [32] and the distinguishable spatial-spectral feature learning neural network method with mean kappa of 0.70 [33] but slightly lower than the 10-fold cross validation result of the CNN-LSTM hybrid method (95.65%) which used window slicing on both EOG and EEG data for data augmentation [34].…”

Section: -Fold Cross Validation On Sessionmentioning

confidence: 84%

FB-CGANet: filter bank channel group attention network for multi-class motor imagery classification

Chen¹,

Yi²,

Wang³

et al. 2022

J. Neural Eng.

View full text Add to dashboard Cite

Objective. Motor imagery-based brain computer interface (MI-BCI) is one of the most important BCI paradigms and can identify the target limb of subjects from the feature of MI-based Electroencephalography (EEG) signals. Deep learning methods, especially lightweight neural networks, provide an efficient technique for MI decoding, but the performance of lightweight neural networks is still limited and need further improving. This paper aimed to design a novel lightweight neural network for improving the performance of multi-class MI decoding. Approach. A hybrid filter bank structure that can extract information in both time and frequency domain was proposed and combined with a novel channel attention method Channel Group Attention (CGA) to build a lightweight neural network Filter Bank Channel Group Attention Network (FB-CGANet). Accompanied with FB-CGANet, the Band Exchange data augmentation method was proposed to generate training data for networks with filter bank structure. Main results. The proposed method can achieve higher 4-class average accuracy (79.4%) than compared methods on the BCI Competition IV IIa dataset in the experiment on the unseen evaluation data. Also, higher average accuracy (93.5%) than compared methods can be obtained in the cross-validation experiment. Significance. This work implies the effectiveness of channel attention and filter bank structure in lightweight neural networks and provides a novel option for multi-class motor imagery classification.

show abstract

Section: -Fold Cross Validation On Sessionmentioning

confidence: 84%

FB-CGANet: filter bank channel group attention network for multi-class motor imagery classification

Chen¹,

Yi²,

Wang³

et al. 2022

J. Neural Eng.

View full text Add to dashboard Cite

show abstract

“…Essentially, the attention mechanism enables the model to focus on the key information, providing an important way to capture more reliable features [54]. Recently, as the attention mechanism has been widely used in the MI-EEG classification tasks [25,55], for example, Liu et al [55] used an attention mechanism on the extracted CSP-wise features and class-wise features to further improve the MI classification accuracy. In our work, the function of the ECA module is to recalibrate features by explicitly modeling the interdependencies between feature channels, and it can be easily applied in the existed CNN model [39,55].…”

Section: Overall Classification Results and Comparisonmentioning

confidence: 99%

“…Recently, as the attention mechanism has been widely used in the MI-EEG classification tasks [25,55], for example, Liu et al [55] used an attention mechanism on the extracted CSP-wise features and class-wise features to further improve the MI classification accuracy. In our work, the function of the ECA module is to recalibrate features by explicitly modeling the interdependencies between feature channels, and it can be easily applied in the existed CNN model [39,55]. Therefore, ECA assign higher weights adaptively on the spectral-temporal features to obtain more discriminative features.…”

Section: Overall Classification Results and Comparisonmentioning

confidence: 99%

A Model Combining Multi Branch Spectral-Temporal CNN, Efficient Channel Attention, and LightGBM for MI-BCI Classification

Jia

Yin

et al. 2023

IEEE Trans. Neural Syst. Rehabil. Eng.

View full text Add to dashboard Cite

Accurately decoding motor imagery (MI) braincomputer interface (BCI) tasks has remained a challenge for both neuroscience research and clinical diagnosis. Unfortunately, less subject information and low signal-to-noise ratio of MI electroencephalography (EEG) signals make it difficult to decode the movement intentions of users. In this study, we proposed an end-to-end deep learning model, a multi-branch spectral-temporal convolutional neural network with channel attention and LightGBM model (MBSTCNN-ECA-LightGBM), to decode MI-EEG tasks. We first constructed a multi branch CNN module to learn spectral-temporal domain features. Subsequently, we added an efficient channel attention mechanism module to obtain more discriminative features. Finally, LightGBM was applied to decode the MI multi-classification tasks. The within-subject cross-session training strategy was used to validate classification results. The experimental results showed that the model achieved an average accuracy of 86% on the two-class MI-BCI data and an average accuracy of 74% on the four-class MI-BCI data, which outperformed current state-of-the-art methods. The proposed MBSTCNN-ECA-LightGBM can efficiently decode the spectral and temporal domain information of EEG, improving the performance of MI-based BCIs.

show abstract

“…The SE module in our method can re-calibrate the data by explicitly modeling interdependencies between EEG channels [23]. As shown in Fig.…”

Section: Discussionmentioning

confidence: 99%

“…The principle of CSP is to seek a spatial projection matrix to maximize the covariance of one class in the EEG whilst simultaneously minimizing the covariance of the other class in the EEG (see equation ( 2)) [22]. For multi-class MI tasks, the one-versus-rest CSP (OVR-CSP) considers one task as one class and the remaining tasks as the other class [23]. Assume the 𝑋 , ∈ 𝑅 × denotes the j-th EEG trial belonging to i-th class, where C and T are the number of channels and timepoints, respectively.…”

Section: A Common Spatial Pattern Features Extraction With Retaining ...mentioning

confidence: 99%

SincNet-Based Hybrid Neural Network for Motor Imagery EEG Decoding

Chang

Jin

Daly

et al. 2022

IEEE Trans. Neural Syst. Rehabil. Eng.

Self Cite

View full text Add to dashboard Cite

It is difficult to identify optimal cut-off frequencies for filters used with the common spatial pattern (CSP) method in motor imagery (MI)-based brain-computer interfaces (BCIs). Most current studies choose filter cut-frequencies based on experience or intuition, resulting in sub-optimal use of MI-related spectral information in the electroencephalography (EEG). To improve information utilization, we propose a SincNet-based hybrid neural network (SHNN) for MI-based BCIs. First, raw EEG is segmented into different time windows and mapped into the CSP feature space. Then, SincNets are used as filter bank band-pass filters to automatically filter the data. Next, we used squeeze-and-excitation modules to learn a sparse representation of the filtered data. The resulting sparse data were fed into convolutional neural networks to learn deep feature representations. Finally, these deep features were fed into a gated recurrent unit module to seek sequential relations, and a fully connected layer was used for classification. We used the BCI competition IV datasets 2a and 2b to verify the effectiveness of our SHNN method. The mean classification accuracies (kappa values) of our SHNN method are 0.7426 (0.6648) on dataset 2a and 0.8349 (0.6697) on dataset 2b, respectively. The statistical test results demonstrate that our SHNN can significantly outperform other state-of-the-art methods on these datasets.

show abstract

Distinguishable spatial-spectral feature learning neural network framework for motor imagery-based brain–computer interface

Cited by 9 publications

References 36 publications

FB-CGANet: filter bank channel group attention network for multi-class motor imagery classification

FB-CGANet: filter bank channel group attention network for multi-class motor imagery classification

A Model Combining Multi Branch Spectral-Temporal CNN, Efficient Channel Attention, and LightGBM for MI-BCI Classification

SincNet-Based Hybrid Neural Network for Motor Imagery EEG Decoding

Contact Info

Product

Resources

About