Residual Network Structure Research Articles

Variable selection of nonparametric spatial autoregressive models via deep learning

With the development of deep learning techniques, the application of neural networks to statistical inference has dramatically increased in popularity. In this paper, we extend the deep neural network-based variable selection method to nonparametric spatial autoregressive models. Our approach incorporates feature selection and parameter learning by introducing Lasso penalties in a residual network structure with spatial effects. We transform the problem into a constrained optimization task, where optimizing an objective function with constraints. Without specifying sparsity, we are also able to obtain a specific set of selected variables. The performance of the method with finite samples is demonstrated through an extensive Monte Carlo simulation study. Finally, we apply the method to California housing price data, further validating its superiority in terms of variable selection and predictive performance.

Lightweight Bearing Fault Diagnosis Method Based on Improved Residual Network

A lightweight bearing fault detection approach based on an improved residual network is presented to solve the shortcomings of previous fault diagnostic methods, such as inadequate feature extraction and an excessive computational cost due to high model complexity. First, the raw data are turned into a time–frequency map using the continuous wavelet transform, which captures all of the signal’s time- and frequency-domain properties. Second, an improved residual network model was built, which incorporates the criss-cross attention mechanism and depth-separable convolution into the residual network structure to realize the important distinction of the extracted features and reduce computational resources while ensuring diagnostic accuracy; simultaneously, the Meta-Acon activation function was introduced to improve the network’s self-adaptive characterization ability. The study findings indicate that the suggested approach had a 99.95% accuracy rate and a floating point computational complexity of 0.53 GF. Compared with other networks, it had greater fault detection accuracy and stronger generalization ability, and it could perform high-precision fault diagnostic jobs due to its lower complexity.

SleepGCN: A transition rule learning model based on Graph Convolutional Network for sleep staging

Background and Objective:Automatic sleep staging is essential for assessing and diagnosing sleep disorders, serving millions of people who suffer from them. Numerous sleep staging models have been proposed recently, but most of them have not fully explored the sleep transition rules that are essential for sleep experts to identify sleep stages. Therefore, one objective of this paper is to develop an automatic sleep staging model to capture the transition rules between sleep stages. Methods:In this paper, we propose a novel sleep staging model named SleepGCN. It utilizes the deep features of electroencephalogram (EEG) and electrooculogram (EOG) signals extracted by the sleep representation learning (SRL) module, in conjunction with the transition rules learned by the sleep transition rule learning (STRL) module to identify sleep stages. Specifically, the SRL module utilizes the residual network (ResNet) and Long Short Term Memory (LSTM) structure to capture the deep time-invariant features and temporal information of each sleep stage from the two-channel EEG-EOG, and then applies a feature enhancement block to obtain the refined features. The STRL module employs a Graph Convolutional Network (GCN) and a transition rule matrix to capture transition rules between sleep stages based on the sequence labels of the input signals. Results:We evaluate SleepGCN on five public datasets: SleepEDF-20, SleepEDF-78, SHHS, DOD-H and DOD-O. Overall, SleepGCN achieves an accuracy of 89.70%, 87.70%, 86.16%, 82.07%, and 81.20%, alongside a macro-average F1-score of 85.20%, 82.70%, 77.69%, 72.44%, and 72.93% across these datasets, respectively. Conclusions:The results achieved by our proposed model are much better than those of all other compared models. The ablation study validates the contributions of the SRL and STRL modules proposed in SleepGCN to the sleep staging tasks. Additionally, it shows that the sleep staging model using two-channel EEG-EOG outperforms the model using single-channel EEG or EOG. Overall, SleepGCN is an effective solution for sleep staging using two-channel EEG-EOG.

An Optimized Object Detection Algorithm for Marine Remote Sensing Images

In order to address the challenge of the small-scale, small-target, and complex scenes often encountered in offshore remote sensing image datasets, this paper employs an interpolation method to achieve super-resolution-assisted target detection. This approach aligns with the logic of popular GANs and generative diffusion networks in terms of super-resolution but is more lightweight. Additionally, the image count is expanded fivefold by supplementing the dataset with DOTA and data augmentation techniques. Framework-wise, based on the Faster R-CNN model, the combination of a residual backbone network and pyramid balancing structure enables our model to adapt to the characteristics of small-target scenarios. Moreover, the attention mechanism, random anchor re-selection strategy, and the strategy of replacing quantization operations with bilinear interpolation further enhance the model’s detection capability at a low cost. Ablation experiments and comparative experiments show that, with a simple backbone, the algorithm in this paper achieves a mAP of 71.2% on the dataset, an improvement in accuracy of about 10% compared to the Faster R-CNN algorithm.

Super-Resolution Image Reconstruction of Wavefront Coding Imaging System Based on Deep Learning Network

Wavefront Coding (WFC) is an innovative technique aimed at extending the depth of focus (DOF) of optics imaging systems. In digital imaging systems, super-resolution digital reconstruction close to the diffraction limit of optical systems has always been a hot research topic. With the design of a point spread function (PSF) generated by a suitably phase mask, WFC could also be used in super-resolution image reconstruction. In this paper, we use a deep learning network combined with WFC as a general framework for images reconstruction, and verify its possibility and effectiveness. Considering the blur and additive noise simultaneously, we proposed three super-resolution image reconstruction procedures utilizing convolutional neural networks (CNN) based on mean square error (MSE) loss, conditional Generative Adversarial Networks (CGAN), and Swin Transformer Networks (SwinIR) based on mean absolute error (MAE) loss. We verified their effectiveness by simulation experiments. A comparison of experimental results shows that the SwinIR deep residual network structure based on MAE loss optimization criteria can generate more realistic super-resolution images with more details. In addition, we used a WFC camera to obtain a resolution test target and real scene images for experiments. Using the resolution test target, we demonstrated that the spatial resolution could be improved from 55.6 lp/mm to 124 lp/mm by the proposed super-resolution reconstruction procedure. The reconstruction results show that the proposed deep learning network model is superior to the traditional method in reconstructing high-frequency details and effectively suppressing noise, with the resolution approaching the diffraction limit.

Precision prediction in grinding processes based on displacement signal conversion into recurrent plot

The processing of grinding data and the prediction of accuracy are extremely complex; this paper proposes a novel prediction method to ensure the machining accuracy of computer numerical control (CNC) grinding machines. It consists of two components, namely the filter decomposition recurrence plot (RP) transformation (FDRP) and the deep inverted residual attention network (DIRAN). A pipeline named FDRP has been designed to address the issues in the processing of data and the shortcomings of RP. Firstly, the displacement signals undergo filtering to preserve essential grinding information while effectively removing noise. Secondly, long-time series signals are decomposed and augmented based on the characteristics of the machining process. Lastly, an RP transformation is applied to one-dimensional time series data, resulting in the generation of images that accurately represent the grinding process. Furthermore, this paper proposes a novel machining accuracy prediction model. The DIRAN uses a multi-layer inverse residual network structure combined with attention mechanism to extract the features of two-dimensional RP, and its performance is better than other typical prediction methods. It can be applied to predict the polar angle of the workpiece in industrial processing and reduce the defect rate.

HR-YOLO: A Multi-Branch Network Model for Helmet Detection Combined with High-Resolution Network and YOLOv5

Automatic detection of safety helmet wearing is significant in ensuring safe production. However, the accuracy of safety helmet detection can be challenged by various factors, such as complex environments, poor lighting conditions and small-sized targets. This paper presents a novel and efficient deep learning framework named High-Resolution You Only Look Once (HR-YOLO) for safety helmet wearing detection. The proposed framework synthesizes safety helmet wearing information from the features of helmet objects and human pose. HR-YOLO can use features from two branches to make the bounding box of suppression predictions more accurate for small targets. Then, to further improve the iterative efficiency and accuracy of the model, we design an optimized residual network structure by using Optimized Powered Stochastic Gradient Descent (OP-SGD). Moreover, a Laplace-Aware Attention Model (LAAM) is designed to make the YOLOv5 decoder pay more attention to the feature information from human pose and suppress interference from irrelevant features, which enhances network representation. Finally, non-maximum suppression voting (PA-NMS voting) is proposed to improve detection accuracy for occluded targets, using pose information to constrain the confidence of bounding boxes and select optimal bounding boxes through a modified voting process. Experimental results demonstrate that the presented safety helmet detection network outperforms other approaches and has practical value in application scenarios. Compared with the other algorithms, the proposed algorithm improves the precision, recall and mAP by 7.27%, 5.46% and 7.3%, on average, respectively.

Surface Defect Detection of Aluminum Profiles Based on Multiscale and Self-Attention Mechanisms.

To address the various challenges in aluminum surface defect detection, such as multiscale intricacies, sensitivity to lighting variations, occlusion, and noise, this study proposes the AluDef-ClassNet model. Firstly, a Gaussian difference pyramid is utilized to capture multiscale image features. Secondly, a self-attention mechanism is introduced to enhance feature representation. Additionally, an improved residual network structure incorporating dilated convolutions is adopted to increase the receptive field, thereby enhancing the network's ability to learn from extensive information. A small-scale dataset of high-quality aluminum surface defect images is acquired using a CCD camera. To better tackle the challenges in surface defect detection, advanced deep learning techniques and data augmentation strategies are employed. To address the difficulty of data labeling, a transfer learning approach based on fine-tuning is utilized, leveraging prior knowledge to enhance the efficiency and accuracy of model training. In dataset testing, our model achieved a classification accuracy of 97.6%, demonstrating significant advantages over other classification models.

Bearing fault diagnosis based on transfer learning with dual-flow manifold ResNet and improved CapsNet

In intelligent fault diagnosis for rolling bearings, the variable operating conditions of rotating machinery lead to inconsistent distributions between training and testing data, inevitably decreasing fault diagnosis accuracy. To overcome this issue, we propose a bearing fault diagnosis method based on transfer learning with a dual-flow manifold residual network and improved capsule network (DMRCN). Firstly, we introduce a dual-flow manifold residual network structure consisting of continuous wide-kernel convolution, continuous narrow-kernel convolution, and manifold learning spatial information description blocks. This architecture is utilized to extract low-level features from the data. Then, we present a novel architecture of the capsule network with self-attention output. We incorporate a self-attention mechanism into the dynamic routing algorithm to transfer richer high-level features from the low-level capsule layer to the higher-level capsule layer. During training, we propose a loss function known as global-local distribution discrepancy loss to enhance the model’s generalization ability and accelerate model convergence. Finally, we validate the performance of DMRCN in cross-condition transfer fault diagnosis on both the Case Western Reserve University public dataset and a laboratory-built bearing experimental rig dataset. Experimental results demonstrate that DMRCN outperforms other commonly used deep transfer learning methods regarding fault diagnosis across different operating conditions.

Attention-Enhanced Dual-Branch Residual Network with Adaptive L-Softmax Loss for Specific Emitter Identification under Low-Signal-to-Noise Ratio Conditions

To address the issue associated with poor accuracy rates for specific emitter identification (SEI) under low signal-to-noise ratio (SNR) conditions, where the single-dimension radar signal characteristics are severely affected by noise, we propose an attention-enhanced dual-branch residual network structure based on the adaptive large-margin Softmax (ALS). Initially, we designed a dual-branch network structure to extract features from one-dimensional intermediate frequency data and two-dimensional time–frequency images, respectively. By assigning different attention weights according to their importance, these features are fused into an enhanced joint feature for further training. This approach enables the model to extract distinctive features across multiple dimensions and achieve good recognition performance even when the signal is affected by noise. In addition, we have introduced L-Softmax to replace the original Softmax and propose the ALS. This approach adaptively calculates the classification margin decision parameter based on the angle between samples and the classification boundary and adjusts the margin values of the sample classification boundaries; it reduces the intra-class distance for the same class while increasing the inter-class distance between different classes without the need for cumbersome experiments to determine the optimal value of decision parameters. Our experimental findings revealed that, in comparison to alternative methods, our proposed approach markedly enhances the model’s capability to extract features from signals and classify them in low-SNR environments, thereby effectively diminishing the influence of noise. Notably, it achieves the highest recognition rate across a range of low-SNR conditions, registering an average increase in recognition rate of 4.8%.

A two-stream conditional generative adversarial network for improving semantic predictions in urban driving scenes

Semantic segmentation is a well-studied topic and one of the most challenging tasks in computer vision applications, such as autonomous driving. Deep learning approaches based on convolutional neural networks (CNN) have demonstrated exceptional success on this task in recent years. Despite this success, existing approaches are plagued by higher-order inconsistencies between the ground truth images and the ones predicted by the segmentation model. This paper proposes a novel post-processing scheme based on adversarial learning to counter these inconsistencies. Such a scheme can be combined with a variety of existing CNN-based semantic segmentation networks to improve their segmentation performances. The proposed scheme is a Two-Stream Conditional Generative Adversarial Network (TScGAN), with one stream having initial semantic segmentation masks predicted by an existing CNN, while the other stream utilizes scene images to retain high-level information under a supervised residual network structure. In addition, TScGAN incorporates a novel dynamic weighting mechanism, which leads to significant and consistent gains in segmentation performance. Several comparative tests on public benchmark driving databases, including Cityscapes, Mapillary, and Berkeley DeepDrive100K, demonstrate the effectiveness of the proposed method when used with state-of-the-art CNN-based semantic segmentation models. Furthermore, the ablation experiment proved the structural rationality of our two-stream structure. The code for TScGAN could be found at https://github.com/epan-utbm/TScGAN-for-Improving-Semantic-Predictions

DNNAM: Image inpainting algorithm via deep neural networks and attention mechanism

Most image inpainting algorithms have problems such as fuzzy images, texture distortion and semantic inaccuracy, and the image inpainting effect is limited when processing photos with large missing sections and resolution levels. The paper proposes an effective image inpainting algorithm via partial multi-scale channel attention mechanism and deep neural networks to address the above phenomenon that existing image inpainting methods using deep learning modules have insufficient perception and representation capabilities for multi-scale features with high proportion of irregular defects. Initially, we used the Res-U-Net module as a generator. The U-Net-like backbone network topology can achieve the encoding and decoding stages of damaged images. Secondly, the residual network structure was built in the encoder and decoder to improve the ability of the proposed network to extract and display the features of the damaged images. Finally, the partial multi-scale channel attention module was inserted in the skip connection with the decoder to increase the efficiency of using the low-level features of the original images. The experimental results of the research can show that the proposed method outperforms state-of-the-art methods in terms of subjective visual perception and objective evaluation indicators on the CelebA, Places2 and Paris Street View datasets.

Prioritizing rear-end crash explanatory factors for injury severity level using deep learning and global sensitivity analysis

Traffic accidents are usually unique events with unpredictable geographical and temporal dimensions; thus, accident injury severity level (INJ-SL) analysis presents formidable categorization and data stability problems. Classical statistical models are limited in their ability to correctly model INJ-SL, whilst sophisticated machine learning approaches do not appear to have any equations to prioritize/analyze multiple contributing factors to forecast accidents accompanying INJ-SLs. In addition, the intercorrelations between the input variables may render the conclusions of a formal sensitivity analysis incorrectly. Rear-end collisions are the most common form of traffic accidents; consequently, their linked INJ-SL requires more research. This paper provides a complex technique based on a deep learning paradigm paired with different indicators of Global Sensitivity Analysis to address all of these concerns. Unlike existing neural network designs, this technique presents a deep residual neural network structure that employs residual shortcuts (i.e., connections). The connections enable the DRNNs to sidestep a few levels of the deep network architecture, evading the regular training with high accuracy issues. Using the trained DRNNs model, a Latin Hypercube sampling simulation was undertaken to determine each explanatory component's influence on the resulting INJ-SL. The latest available data from 2011 to 2018 is used to assess all rear-end collisions in North Carolina. A comparison was made between the performance of two different schemes of data categorization using a set of global sensitivity metrics. It was determined that the devised technique overcame the data heterogeneity problems to achieve an accuracy of 87%. In addition, the proposed sensitivity analysis identified the most relevant factors associated with INJ-SL rear-end collisions.

Residual Convolutional Attention Model With Transfer Learning for Detecting Multianomalous Features in Structural Vibration Data

In response to the data anomalies and frequent false alarms caused by harsh environments in long‐term structural health monitoring (SHM), this study has reframed the detection of abnormal vibration data as a time series classification problem. This approach identifies multiple anomalous features, thereby reducing manual detection costs. The novel developed Convolutional Neural Network with Squeeze‐and‐Excitation and Multi‐Head Self‐Attention (CNN–SE–MHSA) employs a deep residual network structure with channel and spatial attention mechanisms, effectively handling the global long‐term dependencies required for anomaly feature learning. It better understands and utilizes feature information across different levels and dimensions, enhancing classification accuracy in complex anomaly situations. Through t‐SNE dimensionality reduction visualization and interpretability analysis, it is demonstrated that the model excels in identifying critical features. Furthermore, by generating simulated data with a variational autoencoder (VAE) and implementing transfer learning strategies based on these data, the issue of low recognition accuracy for complex anomaly data due to data imbalance can be effectively mitigated. In a 25‐day long‐term monitoring experiment of indoor tunnel lining structures, this method demonstrated an average accuracy rate exceeding 96% and a rapid detection capability within 16 min. The results indicate that this method achieves high accuracy in anomaly detection for long‐term monitoring data, even when relying exclusively on time‐domain data.

Art appreciation model design based on improved PageRank and ECA-ResNeXt50 algorithm.

Image sentiment analysis technology can predict, measure and understand the emotional experience of human beings through images. Aiming at the problem of extracting emotional characteristics in art appreciation, this article puts forward an innovative method. Firstly, the PageRank algorithm is enhanced using tweet content similarity and time factors; secondly, the SE-ResNet network design is used to integrate Efficient Channel Attention (ECA) with the residual network structure, and ResNeXt50 is optimized to enhance the extraction of image sentiment features. Finally, the weight coefficients of overall emotions are dynamically adjusted to select a specific emotion incorporation strategy, resulting in effective bimodal fusion. The proposed model demonstrates exceptional performance in predicting sentiment labels, with maximum classification accuracy reaching 88.20%. The accuracy improvement of 21.34% compared to the traditional deep convolutional neural networks (DCNN) model attests to the effectiveness of this study. This research enriches images and texts' emotion feature extraction capabilities and improves the accuracy of emotion fusion classification.

A Gridless DOA Estimation Method for Sparse Sensor Array

Direction-of-arrival (DOA) estimation is still a pivotal research direction in array signal processing. Traditional algorithms based on the signal subspace and compressed sensing theory usually suffer from off-grid and computational complexity. Deep-learning-based methods usually face difficulty in obtaining labeled datasets. With the development of array technology, sparse sensor arrays can effectively reduce the number of sensors, which in turn reduces the complexity of the hardware. Therefore, effective DOA estimation algorithms for sparse sensor arrays need to be further investigated. An unsupervised deep learning method is proposed here to address the above issues. A training model was built based on the residual network structure. The DOA estimation was implemented using Vandermonde decomposition. Finally, the experimental findings confirmed the efficacy of the proposed algorithms presented in this article.

The use of thematic context-based deep learning in discourse expression of sports news

Sports news is a type of discourse that is characterized by a specific vocabulary, style, and tone, and it is typically focused on conveying information about sporting events, athletes, and teams. Thematic context-based deep learning is a powerful approach that can be used to analyze and interpret various forms of natural language, including the discourse expression of sports news. An application model of sign language and lip language recognition based on deep learning is proposed to facilitate people with hearing impairment to easily obtain sports news content. First, the lip language recognition system is constructed; next, MobileNet lightweight network combined with Long-Short Term Memory (LSTM) is used to extract lip reading features. ResNet-50 residual network structure isadopted to extract the features of sign language; finally, the convergence, accuracy, precision and recall of the model are verified respectively. The results show that the loss of training set and test set converges gradually with the increase of iteration times; the lip language recognition model and the sign language recognition model basically tend to be stable after 14 iterations and 12 iterations, respectively, suggesting a better convergence effect of sign language recognition. The accuracy of sign language recognition and lip language recognition is 98.9% and 87.7%, respectively. In sign language recognition, the recognition accuracy of numbers 1, 2, 4, 6 and 8 can reach 100%. In lip language recognition, the recognition accuracy of numbers 2, 3 and 9 is relatively higher. This exploration can facilitate hearing-impaired people to quickly obtain the relevant content in sports news videos, and also provide help for their communication.

Longitudinal interactions between residual symptoms and physiological stress in the remitted symptom network structure of depression

Residual symptoms and stress are amongst the most reliable predictors of relapse in remitted depression. Standard methodologies often preclude continuous stress sampling or the evaluation of complex symptom interactions. This limits knowledge acquisition relative to the day-to-day interactions between residual symptoms and stress. The study aims to explore the interactions between physiological stress and residual symptoms network structure in remitted depression. Twenty-two individuals remitted from depression completed baseline, daily diary (DD), and post-DD assessments. Self-reported stress and residual symptoms were measured at baseline and post-DD. Daily diaries required participants to use a wearable electrodermal activity (EDA) device during waking hours and complete residual symptom measures twice daily for 3-weeks. Two-step multilevel vector auto-regression models were used to estimate contemporaneous and dynamic networks. Depressed mood and concentration problems were central across networks. Skin conductance responses (SCRs), suicide, appetite, and sleep problems were central in the temporal and energy loss in the contemporaneous network. Increased SCRs predicted decreased energy loss. Residual symptoms and stress showed bi-directional interactions. Overall, depressed mood and concentration problems were consistently central, thus potentially important intervention targets. Non-obtrusive bio-signal measures should be used to provide the clinical evidence-base for modelling the interactions between depressive residual symptoms and stress. Practical implications are discussed throughout related to focusing on symptom-specific interactions in clinical practice, simultaneously reducing residual symptom and stress occurrences, EDA as pioneering signal for stress detection, and the central role of specific residual symptoms in remitted depression.

Recurrent Residual Deformable Conv Unit and Multi-Head with Channel Self-Attention Based on U-Net for Building Extraction from Remote Sensing Images

Considering the challenges associated with accurately identifying building shape features and distinguishing between building and non-building features during the extraction of buildings from remote sensing images using deep learning, we propose a novel method for building extraction based on U-Net, incorporating a recurrent residual deformable convolution unit (RDCU) module and augmented multi-head self-attention (AMSA). By replacing conventional convolution modules with an RDCU, which adopts a deformable convolutional neural network within a residual network structure, the proposed method enhances the module’s capacity to learn intricate details such as building shapes. Furthermore, AMSA is introduced into the skip connection function to enhance feature expression and positions through content–position enhancement operations and content–content enhancement operations. Moreover, AMSA integrates an additional fusion channel attention mechanism to aid in identifying cross-channel feature expression Intersection over Union (IoU) score differences. For the Massachusetts dataset, the proposed method achieves an Intersection over Union (IoU) score of 89.99%, PA (Pixel Accuracy) score of 93.62%, and Recall score of 89.22%. For the WHU Satellite dataset I, the proposed method achieves an IoU score of 86.47%, PA score of 92.45%, and Recall score of 91.62%, For the INRIA dataset, the proposed method achieves an IoU score of 80.47%, PA score of 90.15%, and Recall score of 85.42%.

Improved Faster Region-Based Convolutional Neural Networks (R-CNN) Model Based on Split Attention for the Detection of Safflower Filaments in Natural Environments

The accurate acquisition of safflower filament information is the prerequisite for robotic picking operations. To detect safflower filaments accurately in different illumination, branch and leaf occlusion, and weather conditions, an improved Faster R-CNN model for filaments was proposed. Due to the characteristics of safflower filaments being dense and small in the safflower images, the model selected ResNeSt-101 with residual network structure as the backbone feature extraction network to enhance the expressive power of extracted features. Then, using Region of Interest (ROI) Align improved ROI Pooling to reduce the feature errors caused by double quantization. In addition, employing the partitioning around medoids (PAM) clustering was chosen to optimize the scale and number of initial anchors of the network to improve the detection accuracy of small-sized safflower filaments. The test results showed that the mean Average Precision (mAP) of the improved Faster R-CNN reached 91.49%. Comparing with Faster R-CNN, YOLOv3, YOLOv4, YOLOv5, and YOLOv6, the improved Faster R-CNN increased the mAP by 9.52%, 2.49%, 5.95%, 3.56%, and 1.47%, respectively. The mAP of safflower filaments detection was higher than 91% on a sunny, cloudy, and overcast day, in sunlight, backlight, branch and leaf occlusion, and dense occlusion. The improved Faster R-CNN can accurately realize the detection of safflower filaments in natural environments. It can provide technical support for the recognition of small-sized crops.