Parallel Convolution Research Articles

Model of oil pipeline tiny defects detection based on DDPM gated parallel convolutional swin transformer

Abstract To address the challenges of difficult detection of minute magnetic flux leakage (MFL) defects, insufficient inspection data, and low detection accuracy, the denoising diffusion probabilistic model (DDPM) gate dilated parallel convolution swin transformer (DGPST) is proposed. First, we introduce a DDPM-based data generation model, successfully generating a large quantity of diverse and rich MFL defect samples. Second, a gated parallel convolution layer is introduced into the backbone network. This strategy uses the characteristics of dilated convolution to broaden the receptive field of the model, thus enhancing the integration ability of global information. The addition of gating mechanism enables the model to adjust the calculation of attention weight based on broader context information in advance, which not only complicates the shortcomings of window self-attention in global dependence understanding, but also effectively suppress irrelevant calculation. Finally, the loss function of H Intersection over Union is introduced to improve the mean average precision. Following these enhancements, DGPST attains a satisfactory outcome in detecting tiny defects within the MFL problem. Experimental data indicates the accuracy of the algorithm reaches 95.6% and the delay is reduced to 7.6 ms.

A parallel neural networks for emotion recognition based on EEG signals

Our study proposes a novel Parallel Temporal–Spatial-Frequency Neural Network (PTSFNN) for emotion recognition. The network processes EEG signals in the time, frequency, and spatial domains simultaneously to extract discriminative features. Despite its relatively simple architecture, the proposed model achieves superior performance. Specifically, PTSFNN first applies wavelet transform to the raw EEG signals and then reconstructs the coefficients based on frequency hierarchy, thereby achieving frequency decomposition. Subsequently, the core part of the network performs three independent parallel convolution operations on the decomposed signals, including a novel graph convolutional network. Finally, an attention mechanism-based post-processing operation is designed to effectively enhance feature representation. The features obtained from the three modules are concatenated for classification, with the cross-entropy loss function being adopted. To evaluate the model’s performance, extensive experiments are conducted on the SEED and SEED-IV public datasets. The experimental results demonstrate that PTSFNN achieves excellent performance in emotion recognition tasks, with classification accuracies of 87.63% and 74.96%, respectively. Comparative experiments with previous state-of-the-art methods confirm the superiority of our proposed model, which can efficiently extract emotion information from EEG signals.

PNNGS, a multi-convolutional parallel neural network for genomic selection.

Genomic selection (GS) can accomplish breeding faster than phenotypic selection. Improving prediction accuracy is the key to promoting GS. To improve the GS prediction accuracy and stability, we introduce parallel convolution to deep learning for GS and call it a parallel neural network for genomic selection (PNNGS). In PNNGS, information passes through convolutions of different kernel sizes in parallel. The convolutions in each branch are connected with residuals. Four different Lp loss functions train PNNGS. Through experiments, the optimal number of parallel paths for rice, sunflower, wheat, and maize is found to be 4, 6, 4, and 3, respectively. Phenotype prediction is performed on 24 cases through ridge-regression best linear unbiased prediction (RRBLUP), random forests (RF), support vector regression (SVR), deep neural network genomic prediction (DNNGP), and PNNGS. Serial DNNGP and parallel PNNGS outperform the other three algorithms. On average, PNNGS prediction accuracy is 0.031 larger than DNNGP prediction accuracy, indicating that parallelism can improve the GS model. Plants are divided into clusters through principal component analysis (PCA) and K-means clustering algorithms. The sample sizes of different clusters vary greatly, indicating that this is unbalanced data. Through stratified sampling, the prediction stability and accuracy of PNNGS are improved. When the training samples are reduced in small clusters, the prediction accuracy of PNNGS decreases significantly. Increasing the sample size of small clusters is critical to improving the prediction accuracy of GS.

VSmTrans: A hybrid paradigm integrating self-attention and convolution for 3D medical image segmentation

PurposeVision Transformers recently achieved a competitive performance compared with CNNs due to their excellent capability of learning global representation. However, there are two major challenges when applying them to 3D image segmentation: i) Because of the large size of 3D medical images, comprehensive global information is hard to capture due to the enormous computational costs. ii) Insufficient local inductive bias in Transformers affects the ability to segment detailed features such as ambiguous and subtly defined boundaries. Hence, to apply the Vision Transformer mechanism in the medical image segmentation field, the above challenges need to be overcome adequately. MethodsWe propose a hybrid paradigm, called Variable-Shape Mixed Transformer (VSmTrans), that integrates self-attention and convolution and can enjoy the benefits of free learning of both complex relationships from the self-attention mechanism and the local prior knowledge from convolution. Specifically, we designed a Variable-Shape self-attention mechanism, which can rapidly expand the receptive field without extra computing cost and achieve a good trade-off between global awareness and local details. In addition, the parallel convolution paradigm introduces strong local inductive bias to facilitate the ability to excavate details. Meanwhile, a pair of learnable parameters can automatically adjust the importance of the above two paradigms. Extensive experiments were conducted on two public medical image datasets with different modalities: the AMOS CT dataset and the BraTS2021 MRI dataset. ResultsOur method achieves the best average Dice scores of 88.3 % and 89.7 % on these datasets, which are superior to the previous state-of-the-art Swin Transformer-based and CNN-based architectures. A series of ablation experiments were also conducted to verify the efficiency of the proposed hybrid mechanism and the components and explore the effectiveness of those key parameters in VSmTrans. ConclusionsThe proposed hybrid Transformer-based backbone network for 3D medical image segmentation can tightly integrate self-attention and convolution to exploit the advantages of these two paradigms. The experimental results demonstrate our method's superiority compared to other state-of-the-art methods. The hybrid paradigm seems to be most appropriate to the medical image segmentation field. The ablation experiments also demonstrate that the proposed hybrid mechanism can effectively balance large receptive fields with local inductive biases, resulting in highly accurate segmentation results, especially in capturing details. Our code is available at https://github.com/qingze-bai/VSmTrans.

A 3D boundary-guided hybrid network with convolutions and Transformers for lung tumor segmentation in CT images

—Accurate lung tumor segmentation from Computed Tomography (CT) scans is crucial for lung cancer diagnosis. Since the 2D methods lack the volumetric information of lung CT images, 3D convolution-based and Transformer-based methods have recently been applied in lung tumor segmentation tasks using CT imaging. However, most existing 3D methods cannot effectively collaborate the local patterns learned by convolutions with the global dependencies captured by Transformers, and widely ignore the important boundary information of lung tumors. To tackle these problems, we propose a 3D boundary-guided hybrid network using convolutions and Transformers for lung tumor segmentation, named BGHNet. In BGHNet, we first propose the Hybrid Local-Global Context Aggregation (HLGCA) module with parallel convolution and Transformer branches in the encoding phase. To aggregate local and global contexts in each branch of the HLGCA module, we not only design the Volumetric Cross-Stripe Window Transformer (VCSwin-Transformer) to build the Transformer branch with local inductive biases and large receptive fields, but also design the Volumetric Pyramid Convolution with transformer-based extensions (VPConvNeXt) to build the convolution branch with multi-scale global information. Then, we present a Boundary-Guided Feature Refinement (BGFR) module in the decoding phase, which explicitly leverages the boundary information to refine multi-stage decoding features for better performance. Extensive experiments were conducted on two lung tumor segmentation datasets, including a private dataset (HUST-Lung) and a public benchmark dataset (MSD-Lung). Results show that BGHNet outperforms other state-of-the-art 2D or 3D methods in our experiments, and it exhibits superior generalization performance in both non-contrast and contrast-enhanced CT scans.

NCNet: Deformable medical image registration network based on neighborhood cross-attention combined with multi-resolution constraints

Objective. Existing registration networks based on cross-attention design usually divide the image pairs to be registered into patches for input. The division and merging operations of a series of patches are difficult to maintain the topology of the deformation field and reduce the interpretability of the network. Therefore, our goal is to develop a new network architecture based on a cross-attention mechanism combined with a multi-resolution strategy to improve the accuracy and interpretability of medical image registration. Approach. We propose a new deformable image registration network NCNet based on neighborhood cross-attention combined with multi-resolution strategy. The network structure mainly consists of a multi-resolution feature encoder, a multi-head neighborhood cross-attention module and a registration decoder. The hierarchical feature extraction capability of our encoder is improved by introducing large kernel parallel convolution blocks; the cross-attention module based on neighborhood calculation is used to reduce the impact on the topology of the deformation field and double normalization is used to reduce its computational complexity. Main result. We performed atlas-based registration and inter-subject registration tasks on the public 3D brain magnetic resonance imaging datasets LPBA40 and IXI respectively. Compared with the popular VoxelMorph method, our method improves the average DSC value by 7.9% and 3.6% on LPBA40 and IXI. Compared with the popular TransMorph method, our method improves the average DSC value by 4.9% and 1.3% on LPBA40 and IXI. Significance. We proved the advantages of the neighborhood attention calculation method compared to the window attention calculation method based on partitioning patches, and analyzed the impact of the pyramid feature encoder and double normalization on network performance. This has made a valuable contribution to promoting the further development of medical image registration methods.

Partial coherence enhances parallelized photonic computing

Advancements in optical coherence control1–5 have unlocked many cutting-edge applications, including long-haul communication, light detection and ranging (LiDAR) and optical coherence tomography6–8. Prevailing wisdom suggests that using more coherent light sources leads to enhanced system performance and device functionalities9–11. Our study introduces a photonic convolutional processing system that takes advantage of partially coherent light to boost computing parallelism without substantially sacrificing accuracy, potentially enabling larger-size photonic tensor cores. The reduction of the degree of coherence optimizes bandwidth use in the photonic convolutional processing system. This breakthrough challenges the traditional belief that coherence is essential or even advantageous in integrated photonic accelerators, thereby enabling the use of light sources with less rigorous feedback control and thermal-management requirements for high-throughput photonic computing. Here we demonstrate such a system in two photonic platforms for computing applications: a photonic tensor core using phase-change-material photonic memories that delivers parallel convolution operations to classify the gaits of ten patients with Parkinson’s disease with 92.2% accuracy (92.7% theoretically) and a silicon photonic tensor core with embedded electro-absorption modulators (EAMs) to facilitate 0.108 tera operations per second (TOPS) convolutional processing for classifying the Modified National Institute of Standards and Technology (MNIST) handwritten digits dataset with 92.4% accuracy (95.0% theoretically).

A GAI-based multi-scale convolution and attention mechanism model for music emotion recognition and recommendation from physiological data

Due to the subjectivity of emotions and the limited number of emotion categories, existing deep learning models require assistance to achieve objective, accurate, and flexible personalized music emotion recommendations. This paper introduces a deep learning approach that combines Generative Artificial Intelligence (GAI) and explicitly leverages physiological indicators to enhance the model's intelligence, versatility, and automation. Physiological indicators such as Heart Rate Variability (HRV) and Galvanic Skin Response (GSR) can be measured using sensors placed on the body's surface, providing more precise information about human emotional changes. This research employs a three-dimensional emotion model, including the tension-arousal axis, energy-arousal axis, and valence axis, to explain the correlation and accuracy between music data and emotions. Based on this, a music emotion classifier is designed, incorporating GAI algorithms to recommend music by matching users' physiological and emotional types with the emotional features of music. The classifier uses Mel-Frequency Cepstral Coefficients (MFCC) to transform audio into Mel-spectrogram as input features. The music emotion selection module adopts a GAI framework of Variational Autoencoder (VAE) and integrates multi-scale parallel convolution and attention mechanism modules. Experimental results demonstrate that this approach is competitive compared to existing deep learning architectures on PMEmo, RAVDESS, and Soundtrack datasets. Furthermore, due to GAI's efficient classification capability, this model is suitable for resource-constrained mobile devices and other smart devices. The results of this study can be applied to emotion-based music recommendation systems, contributing to emotional interventions and improving the performance of exercise and music therapy.

An improved deep temporal convolutional network for new energy stock index prediction

Accurate prediction of the stock indexes in the new energy market is of significant importance to both investors and policymakers. However, in response to the volatility and uncertainty characteristic of the new energy market, most scholars currently focus on training prediction methods using features from a single time scale, which cannot capture the fluctuations of new energy stock indexes under different time scales. Therefore, in this paper, a hybrid deep learning model Multi-kernel Parallel TemporalNet (MKP-TemporalNet) is proposed for predicting new energy stock indexes. This model initially incorporates an attention mechanism to calibrate the feature importance of multivariate time series dynamically, and then combines the characteristics of improved Temporal Convolutional Networks (iTCN) and Bidirectional Gated Recurrent Units (BiGRU) to enhance prediction accuracy effectively. In particular, the novelty of the iTCN lies in the development of a multi-kernel parallel convolution structure within a residual layout at the core of the temporal convolution module, to address the low efficiency of traditional TCN's single kernel convolution in extracting temporal features from input sequences at different time scales. Results from evaluating MKP-TemporalNet against several popular machine learning models on six new energy stock indexes confirm its predictive effectiveness in the new energy sector.

Label-aware Attention Network with Multi-scale Boosting for Medical Image Segmentation

Deep medical image segmentation calls for features with strong discrimination and rich scales due to ambiguous background distraction and large variations in object sizes and shapes. In this paper, we propose two modules to obtain these features. First, existing encoders tend to extract similar foreground/background features at blurry boundaries due to mixed-label feature aggregation. To enhance the discrimination of these features, a Label-Aware Attention (LAA) module is presented to reconstruct them by fusing same-label local features. The fusion is guided by local attention maps based on label-aware affinity learning. Second, instead of relying on a single encoder for scale context mining, we propose a Multi-scale Feature Boosting (MFB) module that applies parallel convolution with different receptive fields for scale embedding and integrates an additional backbone for cross-encoder scale reference. Combining LAA and MFB, a new encoder–decoder based framework is presented, where MFBs act as encoder blocks to recursively extract features with rich scale context, while LAA operates in the decoder layer to enhance the label-aware discriminativeness of features. Extensive experiments on three standard medical segmentation datasets demonstrate the effectiveness of the proposed framework.

A lightweight spatially-aware classification model for breast cancer pathology images

Breast cancer is a prevalent malignant tumour with high global incidence. Its diagnosis relies primarily on the analysis of pathological breast images. Owing to the complex organisation of the tumour microenvironment, neural network models are essential as efficient classification tools in the field of pathological image analysis. This study introduced spatially-aware attention swift parallel convolution network (SPA-SPCNet), a lightweight and low-latency model for classifying breast pathologies. A novel module for multi-scale feature extraction was constructed using a depthwise separable convolution method. It focuses on the multi-scale features of pathological images to alleviate recognition problems caused by similar local features in breast cancer tissues. The module concatenates the convolutions of different kernels from three branches. Second, a lightweight dynamic spatially-aware attention module was introduced to integrate the visual graph convolutional architecture in a branch. This allowed the model to capture the spatial structure and relationships in image, enabling better handling of the unique spatial distribution relationship between breast cancer tissue structures. The other branch utilises a self-attention mechanism in the transformer. The module can dynamically adjust the attention of the model to different regions in the image, allowing it to focus on the key features of the complex spatial distribution of breast cancer tissue. This feature fusion method enabled the model to capture both global semantics and local details. Compared with existing lightweight models, the proposed model has advantages in terms of tissue structure classification accuracy, parameter quantity, floating-point operations, and real-time inference speed, providing a powerful tool for computer-aided breast pathological image classification.

TCN-Inception: Temporal Convolutional Network and Inception modules for sensor-based Human Activity Recognition

The field of Human Activity Recognition (HAR) has experienced a significant surge in interest due to its essential role across numerous areas, including human–computer interaction (HCI), healthcare, smart homes, and various Internet of Things (IoT) applications. The power of deep learning methods in performing various classification tasks, including HAR, has been well-demonstrated. In light of this, our paper presents an efficient HAR system developed using a unique deep-learning architecture called TCN-Inception, which is designed for multivariate time series tasks like HAR data, by combining Temporal Convolutional Network (TCN) and Inception modules. The network starts with an Inception module that uses parallel convolution layers with various kernel sizes for feature extraction. It then includes a TCN module with dilated convolutions to grasp extended temporal patterns. Features are merged from different channels, and the use of residual connections and batch normalization improves training and deepens the architecture. We use four public datasets, UCI-HAR, MobiAct, Daphnet, and DSADS to assess the performance of the TCN-Inception model, and it obtains an average accuracy of 96.15%, 98.86%, 92.63%, and 99.50% for each dataset, respectively. Moreover, we compare the TCN-Inception to several deep learning frameworks to verify its performance. Finally, we implement an ablation study using several architectural configurations of the TCN-Inception model.

Ship-Radiated Noise Separation in Underwater Acoustic Environments Using a Deep Time-Domain Network

Ship-radiated noise separation is critical in both military and economic domains. However, due to the complex underwater environments with multiple noise sources and reverberation, separating ship-radiated noise poses a significant challenge. Traditionally, underwater acoustic signal separation has employed blind source separation methods based on independent component analysis. Recently, the separation of underwater acoustic signals has been approached as a deep learning problem. This involves learning the features of ship-radiated noise from training data. This paper introduces a deep time-domain network for ship-radiated noise separation by leveraging the power of parallel dilated convolution and group convolution. The separation layer employs parallel dilated convolution operations with varying expansion factors to better extract low-frequency features from the signal envelope while preserving detailed information. Additionally, we use group convolution to reduce the expansion of network size caused by parallel convolution operations, enabling the network to maintain a smaller size and computational complexity while achieving good separation performance. The proposed approach is demonstrated to be superior to the other common networks in the DeepShip dataset through comprehensive comparisons.

Convolutional neural network with parallel convolution scale attention module and ResCBAM for breast histology image classification

Breast cancer is the most common cause of female morbidity and death worldwide. Compared with other cancers, early detection of breast cancer is more helpful to improve the prognosis of patients. In order to achieve early diagnosis and treatment, clinical treatment requires rapid and accurate diagnosis. Therefore, the development of an automatic detection system for breast cancer suitable for patient imaging is of great significance for assisting clinical treatment. Accurate classification of pathological images plays a key role in computer-aided medical diagnosis and prognosis. However, in the automatic recognition and classification methods of breast cancer pathological images, the scale information, the loss of image information caused by insufficient feature fusion, and the enormous structure of the model may lead to inaccurate or inefficient classification. To minimize the impact, we proposed a lightweight PCSAM-ResCBAM model based on two-stage convolutional neural network. The model included a Parallel Convolution Scale Attention Module network (PCSAM-Net) and a Residual Convolutional Block Attention Module network (ResCBAM-Net). The first-level convolutional network was built through a 4-layer PCSAM module to achieve prediction and classification of patches extracted from images. To optimize the network's ability to represent global features of images, we proposed a tiled feature fusion method to fuse patch features from the same image, and proposed a residual convolutional attention module. Based on the above, the second-level convolutional network was constructed to achieve predictive classification of images. We evaluated the performance of our proposed model on the ICIAR2018 dataset and the BreakHis dataset, respectively. Furthermore, through model ablation studies, we found that scale attention and dilated convolution play an important role in improving model performance. Our proposed model outperforms the existing state-of-the-art models on 200 × and 400 × magnification datasets with a maximum accuracy of 98.74 %.

Broad-UNet-diff: Diffeomorphic Deformable Medical Image Registration based on Multi-scale Feature Learning

Introduction: To propose a medical image registration method with significant performance improvement. The spatial transformation obtained by the traditional deformable image registration technology is not smooth enough, and the calculation amount is too large to solve the optimization parameters. The network model proposed based on deep learning medical image registration technology has some limitations, which cannot guarantee the registration of topological structures, resulting in the loss of spatial features. It makes the model have topological conservation and transform reversibility, has the ability to learn more multi-scale features and complex image structures, and captures finer changes while clearly encoding global correlation. Method: Based on the core UNet model, a deformable image registration method with a new architecture Broad-UNet-diff is proposed. The model is equipped with asymmetric parallel convolution and uses diffeomorphism mapping. Result: Compared with the seven classical registration methods under the brain MRI datasets, the proposed method has significantly improved the registration performance. In particular, compared with the advanced TransMorph-diff registration method, the Dice score can be improved by 12 %, but only the 1/10 parameters are needed. Conclusion: This method confirms its convincing effectiveness and accuracy.

A bearing fault diagnosis method based on M-SSCNN and M-LR attention mechanism

Bearing fault diagnosis is vital for mechanical maintenance and fault prediction. It ensures equipment safety, extends lifespan, reduces maintenance costs, and improves production efficiency. Nevertheless, it should be acknowledged that some existing diagnostic methods have achieved high accuracy rates in certain scenarios. However, the challenge lies in their limited generalization capabilities, which can lead to reduced accuracy when applied to diverse or unseen conditions. In this study, we proposed a new bearing fault diagnosis method to address the issue of low accuracy caused by the inadequate generalization of models in the process of rolling bearing fault diagnosis. The method is based on a multi-scale sliding convolution neural network and multi-level residual attention mechanism, the model exhibits high accuracy, strong generalization capability, and lightweight structure. Firstly, the time domain signal of the bearing vibration is converted into a two-dimensional time–frequency map, and the image is pixel-segmented using superpixel segmentation techniques. Next, a multi-scale parallel convolution approach is used to extract features to improve the adaptability and robustness of the model to objects of different sizes and scales. Sliding convolution is used instead of pooling to avoid the problem of feature loss caused by maximum pooling and average pooling. A multi-level attention mechanism is then introduced for all stacked channels to focus on the more important and critical information of the module, and residual connections are added to prevent degradation of the network performance. Finally, the proposed method is passed through the fully connected layer for classification using the Softmax classifier. Experimental verification using public datasets and experimental data of our research group shows that the proposed method has better performance than the existing diagnostic methods and diagnostic models. The proposed method offers an advanced and innovative solution in the domain of bearing fault diagnosis.

WaveletFormerNet: A Transformer-based wavelet network for real-world non-homogeneous and dense fog removal

Although deep convolutional neural networks have achieved remarkable success in removing synthetic fog, it is essential to be able to process images taken in complex foggy conditions, such as dense or non-homogeneous fog, in the real world. However, the haze distribution in the real world is complex, and downsampling can lead to color distortion or loss of detail in the output results as the resolution of a feature map or image resolution decreases. Moreover, the over-stacking of convolutional blocks might increase the model complexity. In addition to the challenges of obtaining sufficient training data, overfitting can also arise in deep learning techniques for foggy image processing, which can limit the generalization abilities of the model, posing challenges for its practical applications in real-world scenarios. Considering these issues, this paper proposes a Transformer-based wavelet network (WaveletFormerNet) for real-world foggy image recovery. We embed the discrete wavelet transform into the Vision Transformer by proposing the WaveletFormer and IWaveletFormer blocks, aiming to alleviate texture detail loss and color distortion in the image due to downsampling. We introduce parallel convolution in the Transformer block, which allows for the capture of multi-frequency information in a lightweight mechanism. Such a structure reduces computational expenses and improves the effectiveness of the network. Additionally, we have implemented a feature aggregation module (FAM) to maintain image resolution and enhance the feature extraction capacity of our model, further contributing to its impressive performance in real-world foggy image recovery tasks. Through extensive experiments on real-world fog datasets, we have demonstrated that our WaveletFormerNet achieves superior performance compared to state-of-the-art methods, as shown through quantitative and qualitative evaluations of minor model complexity. Additionally, our satisfactory results on real-world dust removal and application tests showcase the superior generalization ability and improved performance of WaveletFormerNet in computer vision-related applications compared to existing state-of-the-art methods, further confirming our proposed approach's effectiveness and robustness. Our code is available at https://github.com/shengli666666/WaveletFormerNet.

Prediction of Sea Surface Temperature Using U-Net Based Model

Sea surface temperature (SST) is a key parameter in ocean hydrology. Currently, existing SST prediction methods fail to fully utilize the potential spatial correlation between variables. To address this challenge, we propose a spatiotenporal UNet (ST-UNet) model based on the UNet model. In particular, in the encoding phase of ST-UNet, we use parallel convolution with different kernel sizes to efficiently extract spatial features, and use ConvLSTM to capture temporal features based on the utilization of spatial features. Atrous Spatial Pyramid Pooling (ASPP) module is placed at the bottleneck of the network to further incorporate the multi-scale features, allowing the spatial features to be fully utilized. The final prediction is then generated in the decoding stage using parallel convolution with different kernel sizes similar to the encoding stage. We conducted a series of experiments on the Bohai Sea and Yellow Sea SST data set, as well as the South China Sea SST data set, using SST data from the past 35 days to predict SST data for 1, 3, and 7 days in the future. The model was trained using data spanning from 2010 to 2021, with data from 2022 being utilized to assess the model’s predictive performance. The experimental results show that the model proposed in this research paper achieves excellent results at different prediction scales in both sea areas, and the model consistently outperforms other methods. Specifically, in the Bohai Sea and Yellow Sea sea areas, when the prediction scales are 1, 3, and 7 days, the MAE of ST-UNet outperforms the best results of the other three compared models by 17%, 12%, and 2%, and the MSE by 16%, 18%, and 9%, respectively. In the South China Sea, when the prediction ranges are 1, 3, and 7 days, the MAE of ST-UNet is 27%, 18%, and 3% higher than the best of the other three compared models, and the MSE is 46%, 39%, and 16% higher, respectively. Our results highlight the effectiveness of the ST-UNet model in capturing spatial correlations and accurately predicting SST. The proposed model is expected to improve marine hydrographic studies.

A two-stage transformer based network for motor imagery classification

Brain-computer interfaces (BCIs) are used to understand brain functioning and develop therapies for neurological and neurodegenerative disorders. Therefore, BCIs are crucial in rehabilitating motor dysfunction and advancing motor imagery applications. For motor imagery, electroencephalogram (EEG) signals are used to classify the subject's intention of moving a body part without actually moving it. This paper presents a two-stage transformer-based architecture that employs handcrafted features and deep learning techniques to enhance the classification performance on benchmarked EEG signals. Stage-1 is built on parallel convolution based EEGNet, multi-head attention, and separable temporal convolution networks for spatiotemporal feature extraction. Further, for enhanced classification, in stage-2, additional features and embeddings extracted from stage-1 are used to train TabNet. In addition, a novel channel cluster swapping data augmentation technique is also developed to handle the issue of limited samples for training deep learning architectures. The developed two-stage architecture offered an average classification accuracy of 88.5 % and 88.3 % on the BCI Competition IV-2a and IV-2b datasets, respectively, which is approximately 3.0 % superior over similar recent reported works.

A deep learning approach for reduce CSI feedback overhead in massive MIMO

The Massive MIMO system in FDD communication mode is still one of the indispensable technologies in future cellular communication systems. In order to reduce CSI feedback overhead and enhance the quality of CSI channel reconstruction, this paper proposes a RiCsiNet network model and designs an iNetBlock module with larger convolutional kernels and a wider network structure. By connecting parallel convolution kernels of 1 × 1, 3 × 3, 5 × 5, and 7 × 7, both the width of the network and the receptive field are increased simultaneously, enabling calculation of different feature details at each layer. The compressed CSI is accurately estimated through residual connections at the BS end. The network model proposed in this paper, when considering both the model parameter size and the estimated CSI accuracy, outperforms the existing algorithm model in the NMSE index of 70%. In this paper, we verify that using different learning rate decay algorithms to train the model, a good learning rate decay algorithm can effectively improve the quality of the model, the best result is 12% higher than the worst result. Furthermore, we demonstrate that training different network models with varying amounts of data can not only save time but also maximize CSI channel estimation quality on the BS side. Finally, the RiCsiNet model was verified using different SNR under COST2100 channel model. QuaDRiGa data sets with different truncated subcarriers are used to verify the RiCsiNet model. The CSI estimated by the RiCsiNet model all showed high accuracy.