A gradient-based automatic optimization CNN framework for EEG state recognition

XCF-LSTMSATNet: A Classification Approach for EEG Signals Evoked by Dynamic Random Dot Stereograms

Stereovision is the visual perception of depth derived from the integration of two slightly different images from each eye, enabling understanding of the three-dimensional space. This capability is deeply intertwined with cognitive brain functions. To explore the impact of stereograms with varied motions on brain activities, we collected Electroencephalography (EEG) signals evoked by Dynamic Random Dot Stereograms (DRDS). To effectively classify the EEG signals induced by DRDS, we introduced a novel hybrid neural network model, XCF-LSTMSATNet, which integrates an XGBoost Channel Feature Optimization Module with the EEGNet and an LSTM Self-Attention Modules. Initially, in the channel selection phase, XGBoost is employed for preliminary classification and feature weight analysis, which can enhance our channel selection strategy. Following this, EEGNet employs deep convolutional layers to extract spatial features, while separable convolutions are subsequently used to derive high-dimensional spatial-temporal features. Meanwhile, the LSTMSAT Module, with its capability to learn long-term dependencies in time-series signals, is deployed to capture temporal continuity information. The incorporation of the self-attention mechanism further amplifies the model's ability to grasp long-distance dependencies and enables dynamic weight allocation to the extracted features. In the end, both temporal and spatial features are integrated into the classification module, enabling precise prediction across three categories of EEG signals. The proposed XCF-LSTMSATNet was extensively tested on both a custom dataset and the public datasets SRDA and SRDB. The results demonstrate that the model exhibits solid classification performance across all three datasets, effectively showcasing its robustness and generalization capabilities.

3D Visual Discomfort Assessment With a Weakly Supervised Graph Convolution Neural Network Based on Inaccurately Labeled EEG

Visual discomfort significantly limits the broader application of stereoscopic display technology. Hence, the accurate assessment of stereoscopic visual discomfort is a crucial topic in this field. Electroencephalography (EEG) data, which can reflect changes in brain activity, have received increasing attention in objective assessment research. However, inaccurately labeled data, resulting from the presence of individual differences, restrict the effectiveness of the widely used supervised learning methods in visual discomfort assessment tasks. Simultaneously, visual discomfort assessment methods should pay greater attention to the information provided by the visual cortical areas of the brain. To tackle these challenges, we need to consider two key aspects: maximizing the utilization of inaccurately labeled data for enhanced learning and integrating information from the brain's visual cortex for feature representation purposes. Therefore, we propose the weakly supervised graph convolution neural network for visual discomfort (WSGCN-VD). In the classification part, a center correction loss serves as a weakly supervised loss, employing a progressive selection strategy to identify accurately labeled data while constraining the involvement of inaccurately labeled data that are influenced by individual differences during the model learning process. In the feature extraction part, a feature graph module pays particular attention to the construction of spatial connections among the channels in the visual regions of the brain and combines them with high-dimensional temporal features to obtain visually dependent spatio-temporal representations. Through extensive experiments conducted in various scenarios, we demonstrate the effectiveness of our proposed model. Further analysis reveals that the proposed model mitigates the impact of inaccurately labeled data on the accuracy of assessment.

Time-Series Anomaly Detection Based on Dynamic Temporal Graph Convolutional Network for Epilepsy Diagnosis

Electroencephalography (EEG) is typical time-series data. Designing an automatic detection model for EEG is of great significance for disease diagnosis. For example, EEG stands as one of the most potent diagnostic tools for epilepsy detection. A myriad of studies have employed EEG to detect and classify epilepsy, yet these investigations harbor certain limitations. Firstly, most existing research concentrates on the labels of sliced EEG signals, neglecting epilepsy labels associated with each time step in the original EEG signal-what we term fine-grained labels. Secondly, a majority of these studies utilize static graphs to depict EEG's spatial characteristics, thereby disregarding the dynamic interplay among EEG channels. Consequently, the efficient nature of EEG structures may not be captured. In response to these challenges, we propose a novel seizure detection and classification framework-the dynamic temporal graph convolutional network (DTGCN). This method is specifically designed to model the interdependencies in temporal and spatial dimensions within EEG signals. The proposed DTGCN model includes a unique seizure attention layer conceived to capture the distribution and diffusion patterns of epilepsy. Additionally, the model incorporates a graph structure learning layer to represent the dynamically evolving graph structure inherent in the data. We rigorously evaluated the proposed DTGCN model using a substantial publicly available dataset, TUSZ, consisting of 5499 EEGs. The subsequent experimental results convincingly demonstrated that the DTGCN model outperformed the existing state-of-the-art methods in terms of efficiency and accuracy for both seizure detection and classification tasks.

STSNet: a novel spatio-temporal-spectral network for subject-independent EEG-based emotion recognition

How to use the characteristics of EEG signals to obtain more complementary and discriminative data representation is an issue in EEG-based emotion recognition. Many studies have tried spatio-temporal or spatio-spectral feature fusion to obtain higher-level representations of EEG data. However, these studies ignored the complementarity between spatial, temporal and spectral domains of EEG signals, thus limiting the classification ability of models. This study proposed an end-to-end network based on ManifoldNet and BiLSTM networks, named STSNet. The STSNet first constructed a 4-D spatio-temporal-spectral data representation and a spatio-temporal data representation based on EEG signals in manifold space. After that, they were fed into the ManifoldNet network and the BiLSTM network respectively to calculate higher-level features and achieve spatio-temporal-spectral feature fusion. Finally, extensive comparative experiments were performed on two public datasets, DEAP and DREAMER, using the subject-independent leave-one-subject-out cross-validation strategy. On the DEAP dataset, the average accuracy of the valence and arousal are 69.38% and 71.88%, respectively; on the DREAMER dataset, the average accuracy of the valence and arousal are 78.26% and 82.37%, respectively. Experimental results show that the STSNet model has good emotion recognition performance.

AutoEER: automatic EEG-based emotion recognition with neural architecture search

Objective.Emotion recognition based on electroencephalography (EEG) is garnering increasing attention among researchers due to its wide-ranging applications and the rise of portable devices. Deep learning-based models have demonstrated impressive progress in EEG-based emotion recognition, thanks to their exceptional feature extraction capabilities. However, the manual design of deep networks is time-consuming and labour-intensive. Moreover, the inherent variability of EEG signals necessitates extensive customization of models, exacerbating these challenges. Neural architecture search (NAS) methods can alleviate the need for excessive manual involvement by automatically discovering the optimal network structure for EEG-based emotion recognition.Approach.In this regard, we propose AutoEER (AutomaticEEG-basedEmotionRecognition), a framework that leverages tailored NAS to automatically discover the optimal network structure for EEG-based emotion recognition. We carefully design a customized search space specifically for EEG signals, incorporating operators that effectively capture both temporal and spatial properties of EEG. Additionally, we employ a novel parameterization strategy to derive the optimal network structure from the proposed search space.Main results.Extensive experimentation on emotion classification tasks using two benchmark datasets, DEAP and SEED, has demonstrated that AutoEER outperforms state-of-the-art manual deep and NAS models. Specifically, compared to the optimal model WangNAS on the accuracy (ACC) metric, AutoEER improves its average accuracy on all datasets by 0.93%. Similarly, compared to the optimal model LiNAS on the F1 Ssore (F1) metric, AutoEER improves its average F1 score on all datasets by 4.51%. Furthermore, the architectures generated by AutoEER exhibit superior transferability compared to alternative methods.Significance.AutoEER represents a novel approach to EEG analysis, utilizing a specialized search space to design models tailored to individual subjects. This approach significantly reduces the labour and time costs associated with manual model construction in EEG research, holding great promise for advancing the field and streamlining research practices.

Improving EEG-Based Driver Distraction Classification Using Brain Connectivity Estimators

This paper discusses a novel approach to an EEG (electroencephalogram)-based driver distraction classification by using brain connectivity estimators as features. Ten healthy volunteers with more than one year of driving experience and an average age of 24.3 participated in a virtual reality environment with two conditions, a simple math problem-solving task and a lane-keeping task to mimic the distracted driving task and a non-distracted driving task, respectively. Independent component analysis (ICA) was conducted on the selected epochs of six selected components relevant to the frontal, central, parietal, occipital, left motor, and right motor areas. Granger-Geweke causality (GGC), directed transfer function (DTF), partial directed coherence (PDC), and generalized partial directed coherence (GPDC) brain connectivity estimators were used to calculate the connectivity matrixes. These connectivity matrixes were used as features to train the support vector machine (SVM) with the radial basis function (RBF) and classify the distracted and non-distracted driving tasks. GGC, DTF, PDC, and GPDC connectivity estimators yielded the classification accuracies of 82.27%, 70.02%, 86.19%, and 80.95%, respectively. Further analysis of the PDC connectivity estimator was conducted to determine the best window to differentiate between the distracted and non-distracted driving tasks. This study suggests that the PDC connectivity estimator can yield better classification accuracy for driver distractions.

Efficient graph convolutional networks for seizure prediction using scalp EEG

Epilepsy is a chronic brain disease that causes persistent and severe damage to the physical and mental health of patients. Daily effective prediction of epileptic seizures is crucial for epilepsy patients especially those with refractory epilepsy. At present, a large number of deep learning algorithms such as Convolutional Neural Networks and Recurrent Neural Networks have been used to predict epileptic seizures and have obtained better performance than traditional machine learning methods. However, these methods usually transform the Electroencephalogram (EEG) signal into a Euclidean grid structure. The conversion suffers from loss of adjacent spatial information, which results in deep learning models requiring more storage and computational consumption in the process of information fusion after information extraction. This study proposes a general Graph Convolutional Networks (GCN) model architecture for predicting seizures to solve the problem of oversized seizure prediction models based on exploring the graph structure of EEG signals. As a graph classification task, the network architecture includes graph convolution layers that extract node features with one-hop neighbors, pooling layers that summarize abstract node features; and fully connected layers that implement classification, resulting in superior prediction performance and smaller network size. The experiment shows that the model has an average sensitivity of 96.51%, an average AUC of 0.92, and a model size of 15.5 k on 18 patients in the CHB-MIT scalp EEG dataset. Compared with traditional deep learning methods, which require a large number of parameters and computational effort and are demanding in terms of storage space and energy consumption, this method is more suitable for implementation on compact, low-power wearable devices as a standard process for building a generic low-consumption graph network model on similar biomedical signals. Furthermore, the edge features of graphs can be used to make a preliminary determination of locations and types of discharge, making it more clinically interpretable.