A robust multi-branch multi-attention-mechanism EEGNet for motor imagery BCI decoding.

BACKGROUND: Motor-Imagery-based Brain-Computer Interface (MI-BCI) is a promising technology to assist communication, movement, and neurological rehabilitation for motor-impaired individuals. Electroencephalography (EEG) decoding techniques using deep learning (DL) possess noteworthy advantages due to automatic feature extraction and end-to-end learning. However, the DL-based EEG decoding models tend to show large variations due to intersubject variability of EEG, which results from inconsistencies of different subjects' optimal hyperparameters. NEW METHODS: This study proposes a multi-branch multi-attention mechanism EEGNet model (MBMANet) for robust decoding. It applies the multi-branch EEGNet structure to achieve various feature extractions. Further, the different attention mechanisms introduced in each branch attain diverse adaptive weight adjustments. This combination of multi-branch and multi-attention mechanisms allows for multi-level feature fusion to provide robust decoding for different subjects. RESULTS: The MBMANet model has a four-classification accuracy of 83.18% and kappa of 0.776 on the BCI Competition IV-2a dataset, which outperforms other eight CNN-based decoding models. This consistently satisfactory performance across all nine subjects indicates that the proposed model is robust. CONCLUSIONS: The combine of multi-branch and multi-attention mechanisms empowers the DL-based models to adaptively learn different EEG features, which provides a feasible solution for dealing with data variability. It also gives the MBMANet model more accurate decoding of motion intentions and lower training costs, thus improving the MI-BCI's utility and robustness.

Personalized motor imagery prediction model based on individual difference of ERP.

Objective. Motor imagery-based brain-computer interaction (MI-BCI) is a novel method of achieving human and external environment interaction that can assist individuals with motor disorders to rehabilitate. However, individual differences limit the utility of the MI-BCI. In this study, a personalized MI prediction model based on the individual difference of event-related potential (ERP) is proposed to solve the MI individual difference.Approach.A novel paradigm named action observation-based multi-delayed matching posture task evokes ERP during a delayed matching posture task phase by retrieving picture stimuli and videos, and generates MI electroencephalogram through action observation and autonomous imagery in an action observation-based motor imagery phase. Based on the correlation between the ERP and MI, a logistic regression-based personalized MI prediction model is built to predict each individual's suitable MI action. 32 subjects conducted the MI task with or without the help of the prediction model to select the MI action. Then classification accuracy of the MI task is used to evaluate the proposed model and three traditional MI methods.Main results.The personalized MI prediction model successfully predicts suitable action among 3 sets of daily actions. Under suitable MI action, the individual's ERP amplitude and event-related desynchronization (ERD) intensity are the largest, which helps to improve the accuracy by 14.25%.Significance.The personalized MI prediction model that uses the temporal ERP features to predict the classification accuracy of MI is feasible for improving the individual's MI-BCI performance, providing a new personalized solution for the individual difference and practical BCI application.

EEGNet-based multi-source domain filter for BCI transfer learning.

Deep learning has great potential on decoding EEG in brain-computer interface. While common deep learning algorithms cannot directly train models with data from multiple individuals because of the inter-individual differences in EEG. Collecting enough data for each subject to satisfy the training of deep learning would result in an increase in training cost. This study proposes a novel transfer learning, EEGNet-based multi-source domain filter for transfer learning (EEGNet-MDFTL), to reduce the amount of training data and improve the performance of BCI. The EEGNet-MDFTL uses bagging ensemble learning to learn domain-invariant features from the multi-source domain and utilizes model loss value to filter the multi-source domain. Compared with baseline methods, the accuracy of the EEGNet-MDFTL reaches 91.96%, higher than two state-of-the-art methods, which demonstrates source domain filter can select similar source domains to improve the accuracy of the model, and remains a high level even when the data amount is reduced to 1/8, proving that ensemble learning learns enough domain invariant features from the multi-source domain to make the model insensitive to data amount. The proposed EEGNet-MDFTL is effective in improving the decoding performance with a small amount of data, which is helpful to save the BCI training cost.

Spatial-temporal variation of extreme precipitation in the Yellow-Huai-Hai-Yangtze Basin of China.

Climate warming leads to frequent extreme precipitation events, which is a prominent manifestation of the variation of the global water cycle. In this study, data from 1842 meteorological stations in the Huang-Huai-Hai-Yangtze River Basin and 7 climate models of CMIP6 were used to obtain the historical and future precipitation data using the Anusplin interpolation, BMA method, and a non-stationary deviation correction technique. The temporal and spatial variations of extreme precipitation in the four basins were analysed from 1960 to 2100. The correlation between extreme precipitation indices and their relationship with geographical factors was also analysed. The result of the study indicates that: (1) in the historical period, CDD and R99pTOT showed an upward trend, with growth rates of 14.14% and 4.78%, respectively. PRCPTOT showed a downward trend, with a decreasing rate of 9.72%. Other indices showed minimal change. (2) Based on SSP1-2.6, the intensity, frequency, and duration of extreme precipitation changed by approximately 5% at SSP3-7.0 and 10% at SSP5-8.5. The sensitivity to climate change was found to be highest in spring and autumn. The drought risk decreased, while the flood risk increased in spring. The drought risk increased in autumn and winter, and the flood risk increased in the alpine climate area of the plateau in summer. (3) Extreme precipitation index is significantly correlated with PRCPTOT in the future period. Different atmospheric circulation factors significantly affected different extreme precipitation indices of FMB. (4) CDD, CWD, R95pD, R99pD, and PRCPTOT are affected by latitude. On the other hand, RX1day and RX5day are affected by longitude. The extreme precipitation index is significantly correlated with geographical factors, and areas above 3000 m above sea level are more sensitive to climate change.

The impact of sodium carbonate on physico-chemical properties and cooking qualities of starches isolated from alkaline yellow noodles.

The physicochemical and cooking properties of wheat starch isolated from alkaline yellow dough treated with sodium carbonate (Na2CO3; 0-3.2 g/100 g) were investigated. With increasing Na2CO3 addition, swelling power increased from 7.28 to 10.70 g/g. X-ray diffraction showed no changes in crystalline patterns while the relative crystallinity decreased from 30.11% to 23.13%. Differential scanning calorimetry results suggested that alkaline salt shifted the gelatinization peak of starch to higher temperatures. The values of pasting viscosity and pasting temperature in alkali-treated starch increased and decreased, respectively. Farinograph results revealed the strengthened structure of dough with alkali-treated starch that was manifested by an increase in the dough development time and dough stability time. Cooking loss and rehydration values of noodles prepared from alkali-treated starch increased by 42% and 36%, respectively. The results suggested that Na2CO3 affected starch crystalline structure, swelling power, gelatinization, pasting properties, starch-gluten interactions and cooking characteristics of noodle products.