This paper proposes a multimodal deep learning framework for automatic sleep stage classification using electroencephalogram (EEG), electrooculogram (EOG), and electrocardiogram (ECG) signals. The model integrates modality-specific convolutional encoders, a cross-attention fusion module, and a bidirectional Transformer for temporal modeling. To alleviate the need for large labeled datasets, a two-stage learning scheme is employed, combining BERT-style self-supervised pre-training with supervised fine-tuning. Experiments on multiple public datasets demonstrate the effectiveness and generalization ability of the framework. On the ISRU C-S3 dataset, it achieves an accuracy of 0.865, an overall F1 score of 0.845, and an Nl-stage F1 of 0.652, surpassing existing methods. These results highlight the model's robustness and potential for real-world deployment.

One of the most prevalent brain-computer interface (BCI) paradigms is the Electroencephalogram (EEG) motor imagery (MI). It has found extensive applications in numerous fields. While there have been significant strides in achieving high MI classification performance, certain challenges still persist:Effective utilization of the time-varying spatial and temporal features from multi-channel brain signals remains elusive. Fully leveraging the interactive information embedded within finite-length MI-EEG samples is still an open question. In this study, we introduce the Deep-Shallow Attention-Based Multi-Frame Fusion Network (DSA-MFNet) tailored for EEG-based motor imagery classification. The architecture of DSA-MFNet encompasses both a Deep-Shallow Attention (DSA) module and a Multi-Frame Fusion (MF) module. In detail, the DSA module integrates a deep-shallow convolution block, which extracts both intricate deep spatial-temporal features and surface-level features. Subsequently, an attention block emphasizes the most salient features in MI-EEG data. Meanwhile, the MF module delves into the interactions amongst multiple frames in the MI-EEG data, shedding light on the unique characteristics of time-varying EEG signals.Our model sets a new benchmark by outperforming the leading techniques, achieving an accuracy of 86.6% on the BCI Competition IV-2a dataset for subject-dependent modes. For the sake of transparency and to foster further research, we will be making our code and trained models available on GitHub.

Deciphering the intricate neural mechanisms behind human speech perception mandates the seamless alignment of auditory signals with electroencephalogram (EEG) data. Yet, the unpredictable fluctuations and inherent noise within EEG, intertwined with the delicate dance between sound and neural response, cast a daunting challenge before researchers. Within the realm of this challenge, our study presents a groundbreaking iteration of the Conformer architecture, meticulously fine-tuned for the exacting task of audio-EEG matching.Where the conventional Conformer pivots on a self-attention mechanism, our enhanced version echoes a more nuanced tune, introducing what we’ve termed "local attention." This deliberate design choice emphasizes nearby inputs, capturing those elusive local patterns and relationships pivotal to understanding the spatial interplay between auditory and EEG features. Such a precision-focused approach guides the model to spotlight vital data sectors during its training voyage, unearthing features rich in meaning and contextual relevance.Our empirical symphony resonates with the brilliance of this enhanced Conformer. It consistently surpasses traditional methods, echoing superior prowess in audio-EEG matching accuracy, and in turn, amplifying the resonance of our innovative approach.

Motor imagery (MI) based on Electroencephalogram (EEG) analysis is a common used paradigm in BCI. Previous works to classify MI have obtained promising classification results. However, there still exit some challenges: 1) Although the deep learning (DL) models are the mainstream methods to solve MI classification, with their depths increasing, the accuracy gets saturated then degrades rapidly. 2) The complex and changeable relationship of the time sequence in the EEG signals makes it difficult to model. In this paper, we propose REEG-BTCNet, a novel framework that achieves outstanding accuracy with stronger robustness for EEG-based motor imagery classification. The REEG-BTCNet consists of residual compact convolution (RCV) module and bi-directional temporal convolution (BTCN) module. Specifically, the RCV module consists of convolution with residual connection to learn high-level task specific EEG feature. The BTCN module consists of a window splitting module and various bi-directional temporal convolutions blocks to model the temporal information from the MI-EEG signals. The proposed model outperforms the current state-of-the-art techniques in the BCI Competition IV-2a dataset with an accuracy of 86.15% for the subject-dependent modes. For reproducibility, the code for this research and the trained models will be released on GitHub.