Simple Recurrent Unit Research Articles

A Light weighted Dense and Tree structured simple recurrent unit (LDTSRU) for flood prediction using meteorological variables

<p style="text-align:justify"><span style="font-size:12pt"><span style="line-height:200%"><span style="font-family:"Times New Roman","serif""><span lang="EN-US" style="line-height:200%">Floods</span><span style="color:black"> cause significant harm around the world each year. Predicting floods accurately and in a </span><span lang="EN-US" style="line-height:200%">timely manner can greatly reduce the loss of human life and property. Thus far, a number of modelling approaches have been described for automatic flood detection; these approaches primarily capture temporal dependences while ignoring patterns related to relative humidity, wind speed, and rainfall intensity—all of which are critical for flood prediction. This paper presents a novel prediction method by combining a Light weighted Dense network and Tree structural simple recurrent unit (LDTSRU). First, a light-weighted dense network is used to convert the input meteorological variables into grayscale images and identify any remarkable patterns between the variables. The nonlinear relationship between the input and output data is then automatically learned using the Tree structural simple recurrent unit (TS-SRU). It is also capable of efficiently comprehending the order and flow of events that culminate in a flood. The p</span><span style="color:black">ublic flood detection dataset is used to confirm the accuracy of the proposed model by comparing it with state-of-the-art methods. According to experimental findings, LDTSRU can perform better with less training time. LDTSRU has attained 2.53% higher average accuracy and better average precision and average recall compared to well-known state-of-the-art techniques.</span></span></span></span></p>

Dynamic community detection algorithm based on hyperbolic graph convolution

PurposeCommunity detection is a key factor in analyzing the structural features of complex networks. However, traditional dynamic community detection methods often fail to effectively solve the problems of deep network information loss and computational complexity in hyperbolic space. To address this challenge, a hyperbolic space-based dynamic graph neural network community detection model (HSDCDM) is proposed.Design/methodology/approachHSDCDM first projects the node features into the hyperbolic space and then utilizes the hyperbolic graph convolution module on the Poincaré and Lorentz models to realize feature fusion and information transfer. In addition, the parallel optimized temporal memory module ensures fast and accurate capture of time domain information over extended periods. Finally, the community clustering module divides the community structure by combining the node characteristics of the space domain and the time domain. To evaluate the performance of HSDCDM, experiments are conducted on both artificial and real datasets.FindingsExperimental results on complex networks demonstrate that HSDCDM significantly enhances the quality of community detection in hierarchical networks. It shows an average improvement of 7.29% in NMI and a 9.07% increase in ARI across datasets compared to traditional methods. For complex networks with non-Euclidean geometric structures, the HSDCDM model incorporating hyperbolic geometry can better handle the discontinuity of the metric space, provides a more compact embedding that preserves the data structure, and offers advantages over methods based on Euclidean geometry methods.Originality/valueThis model aggregates the potential information of nodes in space through manifold-preserving distribution mapping and hyperbolic graph topology modules. Moreover, it optimizes the Simple Recurrent Unit (SRU) on the hyperbolic space Lorentz model to effectively extract time series data in hyperbolic space, thereby enhancing computing efficiency by eliminating the reliance on tangent space.

Application of Unet-SE-Bisru Algorithm for Music Signal Processing in Music Source Separation

At present, the use of time-domain deep learning for end-to-end neural network models has the problem of long training time and poor performance in music source separation. To address this issue, a U-network squeezing excitation bidirectional simple recursive unit model was proposed based on the deep extractor model. Replace Unet SE Bisru with Unet SE Bisru in the following text. This model improves the bidirectional long short-term memory network into a bidirectional simple recurrent unit, and then introduces attention mechanisms in the generalized encoding and decoding layers. The squeezing excitation block is used to selectively extract features based on the type of audio to be separated. Finally, group normalization is added after onedimensional convolution, And its effectiveness was verified. The experimental results show that the signal noise distortion ratio in the improved model is 5.68 decibels compared to the bidirectional simple recursive unit value, which is higher than the 5.55 decibels of bidirectional long short-term memory. After adding the squeezing excitation module, the overall increase is about 0.1-0.5 decibels. In addition, in the model comparison, the three indicators of the improved model with the same number of channels were 5.68 decibels, 5.91 decibels, and 11.28 decibels, respectively, higher than the benchmark model. Compared with other music source classification models, the improved model has better comprehensive separation performance. Although some indicators are lower than the comparison model, the signal noise distortion ratio of drum and bass is 6.11 decibels and 6.36 decibels, which is better than the comparison model. Overall, the improved model has high performance in music source separation for music signal processing and can be effectively applied in practical music source separation.

A data-driven online calibration method for enhancing accuracy in biomedical tactile sensing

Soft tactile sensors have been used to acquire biomedical signals for a variety of applications. Lack of proper calibration can lead to data discrepancies and, in some cases, misdiagnoses. This study introduces a channel-specific calibration technique using an enhanced stacked Simple Recurrent Unit (SRU) network for real-time calibration of each sensor channel. The network is designed to correct deviations due to sensor characteristics or operational pressures through a dynamic weighting mechanism that adjusts to each channel’s energy level. A pneumatic-controlled platform was developed to ensure quality training data for the deep learning model, facilitating a robust public dataset. For pulse wave signals, our approach significantly outperformed existing methods, reducing the Root Mean Square Error (RMSE) by 68.21% over raw signals and 25.68% compared to LSTM-based calibration. It also improved the accuracy of radial augmentation index (radial-AIx) measurements from −3.19% to −0.03%, making biomedical sensors more reliable.

Deep Learning-Driven Public Opinion Analysis on the Weibo Topic about AI Art

The emergence of AI Art has ignited extensive debates on social media platforms. Various online communities have expressed their opinions on different facets of AI Art and participated in discussions with other users, leading to the generation of a substantial volume of data. Analyzing these data can provide useful insights into the public’s opinions on AI Art, enable the investigation of the origins of conflicts in online debates, and contribute to the sustainable development of AI Art. This paper presents a deep learning-driven framework for analyzing the characteristics of public opinion on the Weibo topic of AI Art. To classify the sentiments users expressed in Weibo posts, the linguistic feature-enhanced pre-training model (LERT) was employed to improve text representation via the fusion of syntactic features, followed by a bidirectional Simple Recurrent Unit (SRU) embedded with a soft attention module (BiSRU++) for capturing the long-range dependencies in text features, thus improving the sentiment classification performance. Furthermore, a text clustering analysis was performed across sentiments to capture the nuanced opinions expressed by Weibo users, hence providing useful insights about different online communities. The results indicate that the proposed sentiment analysis model outperforms common baseline models in terms of classification metrics and time efficiency, and the clustering analysis has provided valuable insights for in-depth analyses of AI Art.

PSRUNet: a recurrent neural network for spatiotemporal sequence forecasting based on parallel simple recurrent unit

PSRUNet: a recurrent neural network for spatiotemporal sequence forecasting based on parallel simple recurrent unit

Deep reinforcement learning algorithm based ramp merging decision model

On-ramp merge is a complex traffic scenario in autonomous driving. Because of the uncertainty of the driving environment, most rule-based models cannot solve such a problem. This paper designs a ramp merging decision model based on deep deterministic policy gradient algorithm (DDPG) to solve the vehicle merging problem. To address the problems of slow algorithm merging and poor robustness of previous deep reinforcement learning algorithms in the field of intelligent vehicle ramp merging leading to the low success rate of intelligent vehicle merging, first, we introduce a simple recurrent unit (SRU) for extracting intelligent vehicle states and environment features and use the DDPG algorithm for intelligent vehicle decision making. Second, the experience playback pool of DDPG algorithm is improved by using priority sampling instead of uniform sampling. Finally, a multi-objective reward function is set up during training, considering factors such as safety and efficiency. The simulation experiments show that the improved algorithm improves the merging speed of the model, reduces the collision rate, and enables the vehicle to make more reasonable decisions. In addition, the superiority of the method is demonstrated by comparing with the advanced method.

Ultra-Short-Term Photovoltaic Power Generation Prediction Based on Hunter–Prey Optimized K-Nearest Neighbors and Simple Recurrent Unit

In view of the current problems of complex models and insufficient data processing in ultra-short-term prediction of photovoltaic power generation, this paper proposes a photovoltaic power ultra-short-term prediction model named HPO-KNN-SRU, based on a Simple Recurrent Unit (SRU), K-Nearest Neighbors (KNN), and Hunter–Prey Optimization (HPO). Firstly, the sliding time window is determined by using the autocorrelation function (ACF), partial correlation function (PACF), and model training. The Pearson correlation coefficient method is used to filter the principal meteorological factors that affect photovoltaic power. Then, the K-Nearest Neighbors (KNN) algorithm is utilized for effective outlier detection and processing to ensure the quality of input data for the prediction model, and the Hunter–Prey Optimization (HPO) algorithm is applied to optimize the parameters of the KNN algorithm. Finally, the efficient Simple Recurrent Unit (SRU) model is used for training and prediction, with the Hunter–Prey Optimization (HPO) algorithm applied to optimize the parameters of the SRU model. Simulation experiments and extensive ablation studies using photovoltaic data from the Desert Knowledge Australia Solar Centre (DKASC) in Alice Springs, Australia, validate the effectiveness of the integrated model, the KNN outlier handling, and the HPO algorithm. Compared to the Support Vector Regression (SVR), Long Short-Term Memory (LSTM), Temporal Convolutional Network (TCN), and Simple Recurrent Unit (SRU) models, this model exhibits an average reduction of 19.63% in Mean Square Error (RMSE), 27.54% in Mean Absolute Error (MAE), and an average increase of 1.96% in coefficient of determination (R2) values.

XAI Driven Intelligent IoMT Secure Data Management Framework.

The Internet of Medical Things (IoMT) has transformed traditional healthcare systems by enabling real-time monitoring, remote diagnostics, and data-driven treatment. However, security and privacy remain significant concerns for IoMT adoption due to the sensitive nature of medical data. Therefore, we propose an integrated framework leveraging blockchain and explainable artificial intelligence (XAI) to enable secure, intelligent, and transparent management of IoMT data. First, the traceability and tamper-proof of blockchain are used to realize the secure transaction of IoMT data, transforming the secure transaction of IoMT data into a two-stage Stackelberg game. The dual-chain architecture is used to ensure the security and privacy protection of the transaction. The main-chain manages regular IoMT data transactions, while the side-chain deals with data trading activities aimed at resale. Simultaneously, the perceptual hash technology is used to realize data rights confirmation, which maximally protects the rights and interests of each participant in the transaction. Subsequently, medical time-series data is modeled using bidirectional simple recurrent units to detect anomalies and cyberthreats accurately while overcoming vanishing gradients. Lastly, an adversarial sample generation method based on local interpretable model-agnostic explanations is provided to evaluate, secure, and improve the anomaly detection model, as well as to make it more explainable and resilient to possible adversarial attacks. Simulation results are provided to illustrate the high performance of the integrated secure data management framework leveraging blockchain and XAI, compared with the benchmarks.

Compact deep neural networks for real-time speech enhancement on resource-limited devices

In real-time applications, the aim of speech enhancement (SE) is to achieve optimal performance while ensuring computational efficiency and near-instant outputs. Many deep neural models have achieved optimal performance in terms of speech quality and intelligibility. However, formulating efficient and compact deep neural models for real-time processing on resource-limited devices remains a challenge. This study presents a compact neural model designed in a complex frequency domain for speech enhancement, optimized for resource-limited devices. The proposed model combines convolutional encoder–decoder and recurrent architectures to effectively learn complex mappings from noisy speech for real-time speech enhancement, enabling low-latency causal processing. Recurrent architectures such as Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU), and Simple Recurrent Unit (SRU), are incorporated as bottlenecks to capture temporal dependencies and improve the performance of SE. By representing the speech in the complex frequency domain, the proposed model processes both magnitude and phase information. Further, this study extends the proposed models and incorporates attention-gate-based skip connections, enabling the models to focus on relevant information and dynamically weigh the important features. The results show that the proposed models outperform the recent benchmark models and obtain better speech quality and intelligibility. The proposed models show less computational load and deliver better results. This study uses the WSJ0 database where clean sentences from WSJ0 are mixed with different background noises to create noisy mixtures. The results show that STOI and PESQ are improved by 21.1% and 1.25 (41.5%) on the WSJ0 database whereas, on the VoiceBank+DEMAND database, STOI and PESQ are improved by 4.1% and 1.24 (38.6%) respectively. The extension of the models shows further improvement in STOI and PESQ in seen and unseen noisy conditions.

Short-term prediction of integrated energy load aggregation using a bi-directional simple recurrent unit network with feature-temporal attention mechanism ensemble learning model

In the realm of economic scheduling and optimal operation within integrated energy systems, integrated energy load aggregation plays a pivotal role. Accurate prediction of integrated energy load aggregation enables the refined management of energy resources, thereby enhancing energy utilization efficiency. This paper merges the advantages of clustering algorithms, prediction algorithms and ensemble learning to propose a short-term prediction approach for integrated energy load aggregation. Firstly, the intrinsic distribution of energy consumption data is analyzed through Affinity Propagation (AP) clustering with multi-dimensional similarity. This analysis yields the division of all samples into clusters based on data characteristics. Subsequently, the Transfer entropy is harnessed to conduct correlation analysis within each cluster. For constructing predictive models for each individual cluster, a prediction algorithm based on the Feature-Temporal Attention-enhanced Bi-directional Simple Recurrent Unit (FTA-Bi-SRU) is employed. The Dynamic Weight Average (DWA)-based multi-task loss equilibrium and Multi-Term Adam (MTAdam) loss function are combined to enhance the efficiency of the forecasting model. Finally, a Light Gradient Boosting Machine (LightGBM) model with Bayesian optimization is employed to integrate the predictive outcomes of each individual cluster to generate the overall predictive outcomes for integrated energy load aggregation. The proposed methodology is experimented on real-world energy consumption data to ascertain the efficacy of the prediction method. The experimental results substantiate the superiority of the suggested forecasting approach.

Recurrent ALBERT for recommendation: A hybrid architecture for accurate and lightweight restaurant recommendations

AbstractThe online recommendation system has benefited the traditional restaurant business economically. However, finding the best restaurant during rush time and visiting new places is tough. This objective is addressed through a restaurant recommendation approach, which impacts the human decision‐making method. With the help of collaborative filtering, some user‐based recommendation systems were designed to generate the best recommendation based on user choices. Thus, a user preferences‐based method is presented using A Lite Bidirectional Encoder Representations from Transformers and Simple Recurrent Unit to suggest restaurants based on user preferences. Here, a publicly available dataset from Kaggle called Kzomato is used with 9552 samples and 21 features. And the system obtained an F1‐score, precision, and recall of 86%, which will save time and provide the best recommendation based on user preferences easily.

Attentive Simple Recurrent Unit Knowledge Tracing Based on Learning Ability

Knowledge tracing can analyze the current knowledge level of students through the data of students’ previous learning activities. However, the existing models usually consider the features of exercises, ignoring the individual differences of students. It is difficult to accurately predict students’ mastery. In this paper, we propose an attentive simple recurrent unit knowledge tracing (SRU-MAKT) based on learning ability. The experimental results show that our model is superior to the existing models, and the AUC increases by 1.6%. We also conduct visualization experiments, which show that SRU-MAKT has interpretability.

An Integration of Deep Learning and Transfer Learning for Earthquake-Risk Assessment in the Eurasian Region

The problem of estimating earthquake risk is one of the primary themes for researchers and investigators in the field of geosciences. The combined assessment of spatial probability and the determination of earthquake risk at large scales is challenging. To the best of the authors’ knowledge, there no updated earthquake-hazard-and-risk assessments for the Eurasia region have been published since 1999. Considering that Eurasia is characterized by a seismically active Alpine–Himalayan fault zone and the Pacific Ring of Fire, which are frequently affected by devastating events, a continental-scale risk assessment for Eurasia is necessary to check the global applicability of developed methods and to update the earthquake-hazard, -vulnerability, and -risk maps. The current study proposes an integrated deep-transfer-learning approach called the gated recurrent unit–simple recurrent unit (GRU–SRU) to estimate earthquake risk in Eurasia. In this regard, the GRU model estimates the spatial probability, while the SRU model evaluates the vulnerability. To this end, spatial probability assessment (SPA), and earthquake-vulnerability assessment (EVA) results were integrated to generate risk A, while the earthquake-hazard assessment (EHA) and EVA were considered to generate risk B. This research concludes that in the case of earthquake-risk assessment (ERA), the results obtained for Risk B were better than those for risk A. Using this approach, we also evaluated the stability of the factors and interpreted the interaction values to form a spatial prediction. The accuracy of our proposed integrated approach was examined by means of a comparison between the obtained deep learning (DL)-based results and the maps generated by the Global Earthquake Model (GEM). The accuracy of the SPA was 93.17%, while that of the EVA was 89.33%.

Landslide Susceptibility Mapping Based on Deep Learning Algorithms Using Information Value Analysis Optimization

Selecting samples with non-landslide attributes significantly impacts the deep-learning modeling of landslide susceptibility mapping. This study presents a method of information value analysis in order to optimize the selection of negative samples used for machine learning. Recurrent neural network (RNN) has a memory function, so when using an RNN for landslide susceptibility mapping purposes, the input order of the landslide-influencing factors affects the resulting quality of the model. The information value analysis calculates the landslide-influencing factors, determines the input order of data based on the importance of any specific factor in determining the landslide susceptibility, and improves the prediction potential of recurrent neural networks. The simple recurrent unit (SRU), a newly proposed variant of the recurrent neural network, is characterized by possessing a faster processing speed and currently has less application history in landslide susceptibility mapping. This study used recurrent neural networks optimized by information value analysis for landslide susceptibility mapping in Xinhui District, Jiangmen City, Guangdong Province, China. Four models were constructed: the RNN model with optimized negative sample selection, the SRU model with optimized negative sample selection, the RNN model, and the SRU model. The results show that the RNN model with optimized negative sample selection has the best performance in terms of AUC value (0.9280), followed by the SRU model with optimized negative sample selection (0.9057), the RNN model (0.7277), and the SRU model (0.6355). In addition, several objective measures of accuracy (0.8598), recall (0.8302), F1 score (0.8544), Matthews correlation coefficient (0.7206), and the receiver operating characteristic also show that the RNN model performs the best. Therefore, the information value analysis can be used to optimize negative sample selection in landslide sensitivity mapping in order to improve the model’s performance; second, SRU is a weaker method than RNN in terms of model performance.

Federated-SRUs: A Federated-Simple-Recurrent-Units-Based IDS for Accurate Detection of Cyber Attacks Against IoT-Augmented Industrial Control Systems

The security of Industrial Control Systems (ICS) against cyber-attacks is essential in modern era since ICSs are vital constituent of modern societies and smart cities. However, the augmentation of legacy ICS networks with smart computing and networking technologies (such as, IoT) have intensely enlarged the surface of attacks against these critical infrastructures. This augmentation makes these networks more vulnerable to cyber-attacks and despite the current security solutions, attackers still find ways to proliferate these networks. Intrusion Detection System (IDS) is one of the key security aspect to prevent these networks from contemporary cyber-attacks. Therefore, this paper proposes a new IDS model named Federated-SRUs (simple recurrent units) for the security of IoT-based ICSs. Specifically, the Federated-SRUs IDS model uses an improved simple recurrent units architecture to reduce computational cost and alleviate the gradient vanishing issue in recurrent networks. Then, it performs data aggregation through several communication rounds in federated architecture which allows multiple ICS networks and stakeholders to build a comprehensive IDS model in a privacy-preserving manner. The performance of Federated-SRUs IDS model is validated through experiments using real-world gas pipeline-based ICS network data, which indicates that it is able to accurately detect intrusions in real-time without compromising privacy and security. Experiments also verify that the Federated-SRUs model outperforms existing state-of-the-art approaches and thus can serve as a viable IDS method in IoT-based ICS networks.

A novel multi-step ahead forecasting model for flood based on time residual LSTM

Climate change has significantly impacted hydrology, including extreme precipitation, and changing precipitation patterns that could lead to an increase in flooding. Life and property benefit from accurate and reliable multi-step flood forecasting. Recently, Recurrent Neural Network (RNN) have become increasingly popular among hydrology researchers for their ability to capture historical dependencies, simplify computations by ignoring intermediate hydrological processes, and provide higher prediction accuracy than traditional models. However, RNN-based flood prediction models face two significant challenges. Firstly, due to their strict time-serial, RNN suffer from gradient issues such as vanishing and exploding gradients, which can make training RNN models difficult. To address this issue, we propose a Residual Long Short-Term Memory (ResLSTM) model that incorporates time residual connections into the time connections of LSTM. Secondly, most flood prediction models output a deterministic value, but the natural hydrological characteristics of the basin are a nonlinear and complex system with many influencing factors that have some randomness. This requires the use of probabilistic methods to modeling. Thus, we introduce the probabilistic forecasting model Autoregressive Recurrent Networks (DeepAR) into our flood prediction model, which outputs a prediction interval rather than a deterministic value. Then, we build four flood probability prediction models by combining DeepAR and four enhanced RNN, including ResLSTM (ours), Long Short-Term Memory (LSTM), Gate Recurrent Unit (GRU), and Time Feedforward Connections Simple Gate Recurrent Unit (TFC-SGRU). The performance of these models is evaluated by the long-term hydrologic data of the Passaic and Ramapo River basins in the United States. The results demonstrate that the prediction interval of the four models is more adaptive to flood uncertainties. And the accuracy of peak flow prediction is nearly 100% within a 90% prediction probability interval. The temporal residual-based model is more accurate and robust than the original LSTM and GRU. We believe this study fills a research gap in multi-step-ahead flood probability prediction and improves the accuracy and reliability of flood prediction models.

Video anomaly detection system using deep convolutional and recurrent models

Automatic identification of anomalies in video surveillance is an interesting research field. Even though interactive multimedia anomaly detection algorithms have been developed, it is still hard for video surveillance to find unusual things like illegal activities and crimes. In this study, a deep convolutional neural network (CNN) and a simple recurrent unit (SRU) are used to build an automated system that can find anomalies in videos. The ResNet architecture takes high-level feature representations from the video frames that come in, while the SRU collects temporal features. The SRU has expressive recurrence and allows for highly parallelized implementation, which makes the video anomaly detection system more accurate. In the study, three models to detect anomalies are suggested as ResNet18 + SRU, ResNet34 + SRU, and ResNet50 + SRU, respectively. The suggested models are examined using the UCF-Crime dataset. This study made a clear distinction between normal and unusual actions, showing that CNN + SRU were able to put each unusual action in the right category. Using the UCF-Crime dataset, ResNet18 + SRU achieved 88.92% accuracy, ResNet34 + SRU achieved 89.34% accuracy, and ResNet50 + SRU achieved 91.24% accuracy. Furthermore, the proposed models demonstrated significantly higher performance accuracy and outscored the comparable deep learning models.

SRU-based Multi-angle Enhanced Network for Semantic Text Similarity Calculation of Big Data Language Model

As a fundamental problem of natural language processing (NLP), the calculation of semantic text similarity plays a crucial role in a variety of big data application situations. In the process of text similarity modeling, however, owing to the complexity and ambiguity of Chinese semantics, effectively capturing the semantic interaction characteristics of Chinese text only from a single angle is impossible. This study proposes a deep learning-based computational model for semantic text similarity called SRU-based multi-angle enhanced network (SMAEN). Specifically, the authors firstly combine character-grained embeddings and word-granularity embeddings obtained from the pre-trained model to represent text. The text is encoded using a bidirectional simple recurrent unit (Bi-SRU) network, and the local text similarity is represented using a soft-aligned attention technique. In addition, the authors integrate Bi-SRU with an improved convolutional neural network (CNN) for global similarity modeling to capture semantic, time, and spatial characteristics of short text interaction. Finally, they employ a pooling layer to aggregate the calculation results into a fixed-length vector and a multi-layer perceptual (MLP) classifier to make a determination. Experimental results on Chinese public datasets LCQMC and PAWS-X show that the proposed method fully captures semantic interaction features from multiple angles and achieves advanced performance. This method can produce better matching results and enhance the accuracy of large data analysis. It is applicable to numerous scenarios involving large data, such as information retrieval and recommendation systems.

Sentiment recognition model of EEG signals combined with one-dimensional convolution and BiBASRU-AT

To address the problems of low accuracy of EEG emotion sentiment and insufficient feature extraction ability of recurrent model, an EEG sentiment recognition model combining one-dimensional convolution and BiBASRU-AT is proposed. The data set is preprocessed in segments to expand the number of samples, and 62 channel local emotional features are extracted from one-dimensional convolution; The built-in self-attention simple recurrent unit is constructed to capture the multi-channel fusion features and the dependence between channels. The soft attention mechanism identifies the key features that have a great impact on the identification of emotional tendencies, and the linear layer outputs the positive, neutral and negative emotion recognition results. The experimental results on the EEG data set(SEED) show that the model achieves an average classification accuracy of 90.24%, which is higher than the excellent deep learning model compared with the experiment. The built-in self-attention simple recurrent unit has stronger feature capture ability, which proves the effectiveness of the model.