Human emotions are intricate psychological phenomena that reflect an individual’s current physiological and psychological state.Emotions have a pronounced influence on human behavior,cognition,communication,and decis...Human emotions are intricate psychological phenomena that reflect an individual’s current physiological and psychological state.Emotions have a pronounced influence on human behavior,cognition,communication,and decision-making.However,current emotion recognition methods often suffer from suboptimal performance and limited scalability in practical applications.To solve this problem,a novel electroencephalogram(EEG)emotion recognition network named VG-DOCoT is proposed,which is based on depthwise over-parameterized convolutional(DO-Conv),transformer,and variational automatic encoder-generative adversarial network(VAE-GAN)structures.Specifically,the differential entropy(DE)can be extracted from EEG signals to create mappings into the temporal,spatial,and frequency information in preprocessing.To enhance the training data,VAE-GAN is employed for data augmentation.A novel convolution module DO-Conv is used to replace the traditional convolution layer to improve the network.A transformer structure is introduced into the network framework to reveal the global dependencies from EEG signals.Using the proposed model,a binary classification on the DEAP dataset is carried out,which achieves an accuracy of 92.52%for arousal and 92.27%for valence.Next,a ternary classification is conducted on SEED,which classifies neutral,positive,and negative emotions;an impressive average prediction accuracy of 93.77%is obtained.The proposed method significantly improves the accuracy for EEG-based emotion recognition.展开更多
With the remarkable success of change detection(CD)in remote sensing images in the context of deep learning,many convolutional neural network(CNN)based methods have been proposed.In the current research,to obtain a be...With the remarkable success of change detection(CD)in remote sensing images in the context of deep learning,many convolutional neural network(CNN)based methods have been proposed.In the current research,to obtain a better context modeling method for remote sensing images and to capture more spatiotemporal characteristics,several attention-based methods and transformer(TR)-based methods have been proposed.Recent research has also continued to innovate on TR-based methods,and many new methods have been proposed.Most of them require a huge number of calculation to achieve good results.Therefore,using the TR-based mehtod while maintaining the overhead low is a problem to be solved.Here,we propose a GNN-based multi-scale transformer siamese network for remote sensing image change detection(GMTS)that maintains a low network overhead while effectively modeling context in the spatiotemporal domain.We also design a novel hybrid backbone to extract features.Compared with the current CNN backbone,our backbone network has a lower overhead and achieves better results.Further,we use high/low frequency(HiLo)attention to extract more detailed local features and the multi-scale pooling pyramid transformer(MPPT)module to focus on more global features respectively.Finally,we leverage the context modeling capabilities of TR in the spatiotemporal domain to optimize the extracted features.We have a relatively low number of parameters compared to that required by current TR-based methods and achieve a good effect improvement,which provides a good balance between efficiency and performance.展开更多
基金supported by the National Key Research and Development Program of China(No.2022YFE0122700)the National Natural Science Foundation of China(No.61971230)。
文摘Human emotions are intricate psychological phenomena that reflect an individual’s current physiological and psychological state.Emotions have a pronounced influence on human behavior,cognition,communication,and decision-making.However,current emotion recognition methods often suffer from suboptimal performance and limited scalability in practical applications.To solve this problem,a novel electroencephalogram(EEG)emotion recognition network named VG-DOCoT is proposed,which is based on depthwise over-parameterized convolutional(DO-Conv),transformer,and variational automatic encoder-generative adversarial network(VAE-GAN)structures.Specifically,the differential entropy(DE)can be extracted from EEG signals to create mappings into the temporal,spatial,and frequency information in preprocessing.To enhance the training data,VAE-GAN is employed for data augmentation.A novel convolution module DO-Conv is used to replace the traditional convolution layer to improve the network.A transformer structure is introduced into the network framework to reveal the global dependencies from EEG signals.Using the proposed model,a binary classification on the DEAP dataset is carried out,which achieves an accuracy of 92.52%for arousal and 92.27%for valence.Next,a ternary classification is conducted on SEED,which classifies neutral,positive,and negative emotions;an impressive average prediction accuracy of 93.77%is obtained.The proposed method significantly improves the accuracy for EEG-based emotion recognition.
基金The authors acknowledge the National Natural Science Foundation of China(Grant nos.61772319,62002200,62202268 and 62272281)Shandong Natural Science Foundation of China(Grant no.ZR2021QF134 and ZR2021MF107)Yantai Science And Technology Innovation Development Plan(2022JCYJ031).
文摘With the remarkable success of change detection(CD)in remote sensing images in the context of deep learning,many convolutional neural network(CNN)based methods have been proposed.In the current research,to obtain a better context modeling method for remote sensing images and to capture more spatiotemporal characteristics,several attention-based methods and transformer(TR)-based methods have been proposed.Recent research has also continued to innovate on TR-based methods,and many new methods have been proposed.Most of them require a huge number of calculation to achieve good results.Therefore,using the TR-based mehtod while maintaining the overhead low is a problem to be solved.Here,we propose a GNN-based multi-scale transformer siamese network for remote sensing image change detection(GMTS)that maintains a low network overhead while effectively modeling context in the spatiotemporal domain.We also design a novel hybrid backbone to extract features.Compared with the current CNN backbone,our backbone network has a lower overhead and achieves better results.Further,we use high/low frequency(HiLo)attention to extract more detailed local features and the multi-scale pooling pyramid transformer(MPPT)module to focus on more global features respectively.Finally,we leverage the context modeling capabilities of TR in the spatiotemporal domain to optimize the extracted features.We have a relatively low number of parameters compared to that required by current TR-based methods and achieve a good effect improvement,which provides a good balance between efficiency and performance.
