A composite improved attention convolutional network for motor imagery EEG classification

Frontiers in Neuroscience, Journal Year: 2025, Volume and Issue: 19

Published: Feb. 5, 2025

A brain-computer interface (BCI) is an emerging technology that aims to establish a direct communication pathway between the human brain and external devices. Motor imagery electroencephalography (MI-EEG) signals are analyzed infer users' intentions during motor imagery. These hold potential for applications in rehabilitation training device control. However, classification accuracy of MI-EEG remains key challenge development BCI technology. This paper proposes composite improved attention convolutional network (CIACNet) classification. CIACNet utilizes dual-branch neural (CNN) extract rich temporal features, block module (CBAM) enhance feature extraction, (TCN) capture advanced multi-level concatenation more comprehensive representation. The model performs well on both IV-2a IV-2b datasets, achieving accuracies 85.15 90.05%, respectively, with kappa score 0.80 datasets. results indicate model's performance exceeds four other comparative models. Experimental demonstrate proposed has strong capabilities low time cost. Removing one or blocks decline overall model, indicating each within makes significant contribution its effectiveness. ability reduce costs improve (MI-BCI) systems, while also highlighting practical applicability.

Language: Английский

---

Dual-pathway EEG model with channel attention for virtual reality motion sickness detection

Journal of Neuroscience Methods, Journal Year: 2025, Volume and Issue: unknown, P. 110425 - 110425

Published: March 1, 2025

Language: Английский

---

Scaled Custom Attention for Enhanced Temporal Dependency Modeling in EEG Classification

Research Square (Research Square), Journal Year: 2025, Volume and Issue: unknown

Published: April 2, 2025

Accurate Electroencephalography (EEG) classification is essential for diagnosing brain disorders such as Epilepsy. Whereas Deep Learning models Convolution Neural Networks (CNNs) and Long Short Term Memory (LSTM) improved EEG performance over traditional methods, existing attention mechanisms Additive, Luong Multi-head struggle to capture EEG’s complex temporal dependencies. This study proposes Scaled Custom Attention (SCA); a mechanism dependency modeling during classification. Unlike Query-Key-Value (QKV) approaches which rely on semantic weighting schemes, SCA employs direct feature strategy that adapts the unique dependencies of signals, introduces scaling enhances stability. To validate our approach, experiments were conducted using TUH Epilepsy Corpus (TUEP) where achieved compelling accuracy (98.17%), surpassing Additive (96.47%), Multihead (97.65%), (97.26%) when integrated LConvNet model. Additionally, demonstrated strong scalability, parameter efficiency, generalization abilities, making it promising enhancement EEG-based deep learning models.

Language: Английский

---

Enhancing motor imagery classification: a novel CNN with self-attention using local and global features of filtered EEG data

Connection Science, Journal Year: 2024, Volume and Issue: 36(1)

Published: Nov. 16, 2024

Language: Английский

---

Periodicity-based Multi-Dimensional Interaction Convolution Network with Multi-Scale Feature Fusion for Motor Imagery EEG classification

Journal of Neuroscience Methods, Journal Year: 2024, Volume and Issue: unknown, P. 110356 - 110356

Published: Dec. 1, 2024

Language: Английский

---