Speech emotion recognition (SER), a rapidly evolving task that aims to recognize the emotion of speakers, has become a key research area in affective computing. However, various languages in multilingual natural scenarios extremely challenge the generalization ability of SER, causing the model performance to decrease quickly, and driving researchers to ask how to improve the performance of multilingual SER. Recent studies mainly use feature fusion and language-controlled models to address this challenge, but key points such as the intrinsic association of languages or deep analysis of multilingual shared features (MSFs) are still neglected. To solve this problem, an explainable Multitask-based Shared Feature Learning (MSFL) model is proposed for multilingual SER. The introduction of multi-task learning (MTL) can provide related task information of language recognition for MSFL, improve its generalization in multilingual situations, and further lay the foundation for learning MSFs. Specifically, considering the generalization capability and interpretability of the model, the powerful MTL module was combined with the long short-term memory and attention mechanism, aiming to maintain the generalization in multilingual situations. Then, the feature weights acquired from the attention mechanism were ranked in descending order, and the top-ranked MSFs were compared with top-ranked monolingual features, enhancing the model interpretability based on the feature comparison. Various experiments were conducted on Emo-DB, CASIA, and SAVEE corpora from the model generalization and interpretability aspects. Experimental results indicate that MSFL performs better than most state-of-the-art models, with an average improvement of 3.37–4.49%. Besides, the top 10 features in MSFs almost contain the top-ranked features in three monolingual features, which effectively demonstrates the interpretability of MSFL.

This paper investigated the interregional correlation changed by sport training through electroencephalography (EEG) signals using the techniques of classification and feature selection. The EEG data are obtained from students with long-time professional sport training and normal students without sport training as baseline. Every channel of the 19-channel EEG signals is considered as a node in the brain network and Pearson Correlation Coefficients are calculated between every two nodes as the new features of EEG signals. Then, the Partial Least Square (PLS) is used to select the top 10 most varied features and Pearson Correlation Coefficients of selected features are compared to show the difference of two groups. Result shows that the classification accuracy of two groups is improved from 88.13% by the method using measurement of EEG overall energy to 97.19% by the method using EEG correlation measurement. Furthermore, the features selected reveal that the most important interregional EEG correlation changed by training is the correlation between left inferior frontal and left middle temporal with a decreased value.

Editorial: Neural Dynamics – Models and Complexity

Based upon the command and control network infrastructure of the Army, now electronic military systems are being built to store, transfer, query and manipulate data for various military tasks. However, in the open and interconnected network environment, military systems are facing two main typical security threats, i.e., broken authentication and broken access control. This paper proposes an idea of building a security Web-based electronic military system by utilizing network security trust units. An authorization component based on PKI (Public Key Infrastructure) and LDAP (Lightweight Directory Access Protocol), as well as an access control component based on T-RBAC (Task-Role Based Access Control) are designed. The implemented security components are embedded into an online manuscript submission and editing system for CAPF (Chinese Armed Police Force) newspaper office to ensure data and workflow security. Practical application shows the implemented security components can effectively guarantee the security and reliability of the system.

To explore the effective method to detect the effect of exercise training on EEG activity, this paper compared the distinguishability of EEG changes relevanted to exercise training in different cognitive states. EEG signals of college students who have undergone long-term exercise training (exercise training group) and college students who have not undergone such training (control group) were collected under specific cognitive tasks (Motor imagination tasks). Then the power characteristics of alpha wave and beta wave were obtained by wavelet transform method. The machine learning method was used to distinguish the difference of brain electrical activity between the exercise-trained subjects and the control group during the exercise-related cognitive tasks. Results showed that EEG patterns between two groups were not distinguished in rest status (the accuracy is lower than 59%), but they were separable in the completion of the motor cognitive tasks. The accuracy was over 85%. The pattern of brain electrical activity during cognitive tasks can better reflect the effect of exercise training.