.png){kind=icon}
Return
MDMS: A Multidomain Multiscale Feature Extraction Method for Brain–ComputerInterface
.png){kind=icon}
PRE
AI
DOI：10.1109/JIOT.2026.3661616.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Attention deficit disorders have been confirmed by numerous psychological studies to be a critical component of many mental diseases. Electroencephalography (EEG)-based brain–computer interfaces (BCIs) have been widely focused and extensively studied in recent years for the prospective application in attention recognition. Recently, deep learning methods have been extensively introduced and developed in the BCI field and have achieved good performance. However, most existing deep learning methods fail to fully exploit spectral domain information, which is a pivotal feature and is widely used in traditional machine learning methods. Inspired by this, a multidomain multiscale (MDMS) feature extraction method is proposed to realize attention state classification. Specifically, 2-D EEG signals are extended into three dimensions by a continuous wavelet transform to obtain spectral domain signals. To obtain multidomain multiscale features, convolution kernels with various sizes are utilized to extract temporal, spectral, and spatial features in three dimensions. However, there are cross-scale dependencies, which can result in redundancy and therefore influence the effect of the network. To capture the cross-scale dependencies and reduce the redundancy, a Transformer-based feature fusion block is employed after obtaining multidomain features. To validate the performance of the proposed method, it is compared with several existing methods on two open datasets. The proposed method outperforms the baseline methods with accuracies of 88.36% $\pm ~5.80$ % and 85.79% $\pm ~5.80$ %, which verifies the effectiveness.
Keywords:
Attention state classification
brain–computer interfaces (BCI)
cognitive task
electroencephalography (EEG)
Journal
IF:
8.9
Papers:
1.4W
Citations:
7.8W
Organization
Cited Papers
ShareSaveShareSaveSustaining attention for a prolonged period of time increases temporal variability in cortical responses.png)
CORTEX
OAAI
CORTEXIF3.3
OAAIShareSaveEOG biofeedback protocol based on selecting distinctive features to treat or reduce ADHD symptoms
PREAI
ShareSaveEnhancing Local Climate Zone Classification With MSCA-MSLCZNet: A Multistream Deep Learning Approach
IF0
PREAI
ShareSave
View more