Urban broadband networks are crucial for communication, video streaming, and other digital services. However, in today's world, where high-speed connections are a necessity, studying network reliability, in particular, packet loss, has become vital. This urban broadband packet loss is the focus of this measurement-based study. Using urban monitoring probes, we gather packet data from multiple access networks, including fiber, DSL, and wireless broadband. Our goal is to determine the location and time packet loss occurs, its dependence on network traffic, and its response to varying time-of-day conditions. To quantify the loss burst and the distribution burst, a statistical modeling technique is used. We study the effects of packet loss on end-user services like VoIP and video streaming and thus calculate the practical Quality of Experience (QoE). We also propose a classifier for loss patterns in networks, which draws temporal feature vectors and uses supervised learning for validation. Our analysis shows that the majority of packet losses are bursty and linked to certain congested times. This research can help improve construction planning, resource distribution, and flexible network system algorithms.

Abstract Human Activity Recognition (HAR) holds significant importance in health and human-machine interaction. However, recognizing actions from 2D information faces challenges like occlusion, illumination variation, cluttered backgrounds, and view invariance. These hurdles are particularly pronounced in indoor patient monitoring settings due to fluctuating lighting conditions and cluttered backgrounds, which compromise the accuracy of activity recognition systems. A new architecture named IlluminationRevive has been proposed to tackle this issue, which utilizes an encoder-decoder convolutional neural network (CNN) and image post-processing blocks to enhance the image’s visual appearance. A new dataset comprising seven indoor physical activities has been proposed, created with contributions from thirty individuals aged 20–45. A hybrid fusion architecture is proposed to classify activities, integrating motion sequence information and body joint features. The proposed classification model incorporates generated Skeleton Motion History Images (SMHIs), collected human joint motion features from video frames, and novel kinematic and geometric features within window frames as inputs. The model can extract spatial and temporal feature vectors by integrating ResNet50-ViT (Residual Network-50 layers, Vision Transformer) and CNN-BiLSTM (Convolutional Neural Network-Bidirectional Long Short-Term Memory) layers. The suggested classification model was evaluated alongside state-of-the-art models using the LNMIIT-SMD (The LNM Institute of Information Technology-Skeleton Motion Dataset) and established NTU-RGBD (Nanyang Technological University’s Red Blue Green and Depth information) dataset. The evaluation aimed to assess the effectiveness of the proposed classification model architecture. Results demonstrate the superiority of the proposed model, achieving impressive accuracies of 98.21% on real-time data, 98.45% on the proposed dataset, and 97.12% on the NTU-RGBD dataset. This high-accuracy, low-latency approach enhances robotic perception for healthcare applications, enabling service robots to perform real-time patient monitoring and assistive tasks in dynamic indoor environments.

Human Activity Recognition (HAR) holds significant importance in health and human-machine interaction. However, recognizing actions from 2D information faces challenges like occlusion, illumination variation, cluttered backgrounds, and view invariance. These hurdles are particularly pronounced in indoor patient monitoring settings due to fluctuating lighting conditions and cluttered backgrounds, which compromise the accuracy of activity recognition systems. A new architecture named illuminationRevive has been proposed to tackle this issue. Which utilizes an encoder-decoder convolutional neural network and image post-processing blocks to enhance the image's visual appearance. A new dataset comprising seven indoor physical activities has been proposed, created with contributions from thirty individuals aged 20 to 45. A hybrid fusion architecture is proposed to classify activities, integrating motion sequence information and body joint features. The proposed classification model incorporates generated Skeleton Motion History Images (SMHIs), collected human joint motion features from video frames, and novel kinematic and geometric features within window frames as inputs. The model can extract spatial and temporal feature vectors by integrating ResNet50-ViT (Residual Network-50 layers, Vision Transformer) and CNN-BiLSTM (Convolutional Neural Network-Bidirectional Long Short-Term Memory) layers. The suggested classification model was evaluated alongside state-of-the-art models using the LNMIIT-SMD (The LNM Institute of Information Technology-Skelton Motion Dataset) and established NTU-RGBD (Nanyang Technological University's Red Blue Green and Depth information) dataset. The evaluation aimed to assess the effectiveness of the proposed classification model architecture. Results demonstrate the superiority of the proposed model, achieving impressive accuracies of 98.21% on real-time data, 98.45% on the proposed dataset, and 97.12% on the NTU-RGBD dataset.

Abstract Effectively leveraging the spatial features of time series signals to improve the accuracy of bearing fault classification in neural networks presents a significant challenge. To address this issue of different operating conditions, a novel model termed spatial pyramid pooling residual network-deep belief network (SPRout-DBN) is proposed. First and foremost, the Gramian angular difference fields (GADF) are utilized to encode original vibration signals of bearings. Secondly, two-dimensional images transformed by GADF from original signals are input to a novel designed residual network with spatial pyramid pooling to extract fixed-size temporal fusion feature vectors. Finally, a deep belief network is employed for classification and cross-domain learning, enabling the identification of fault samples under varying operating conditions. The proposed method is validated by two sets of datasets from Case Western Reserve University and Jiangnan University, achieving accuracies of 99.81% and 99.0% under identical operating conditions, and 99.41% and 98.43% under different operating conditions with 40 samples. Comparative analysis indicates that the proposed SPRout-DBN remains more robust and effective compared with other methods such as K-nearest neighbors, support vector machines, LeNet-5, ResNet-18, domain adaptation networks, and domain-adversarial neural networks in diverse operating environments.

In order to enhance the driving safety of intelligent vehicles in complex road scenarios, a method for vehicle operation risk assessment and early warning based on the predictive risk field is proposed. The temporal feature vector composed of the spatiotemporal state characteristics of the ego vehicle and surrounding traffic participants is taken as input data for the Attention‐Bidirectional Long‐Short Term Memory (Attention‐BiLSTM) model, which is trained to establish the desired mapping relationship. By predicting the motion state of the target vehicle and utilizing an improved risk field model based on the target vehicle of heading angle, the predictive risk field is obtained. This allows for the assessment of the ego vehicle operational risks. The risk warning model is integrated to provide risk early warning, and the safety path for the ego vehicle is planned based on the interaction between the predictive risk field equipotential lines and the cubic spline curves. Experimental results demonstrate that the proposed vehicle operation risk assessment and early warning model is effective in providing early warnings and safe path references for the ego vehicle in complex urban road test scenarios.

Abstract In this paper, we first obtain the most representative subset of learning behavior features from big data and evaluate each feature value using a genetic algorithm to obtain its weight parameter. Then a similar least squares method is used to establish behavioral performance indicators through temporality, and the loss function is used to classify the indicators to obtain accurate teaching parameters. Finally, the K-means algorithm is used to mine the spacing distance of different log data and obtain the corresponding temporal feature vector to the learning behavior data with similarity. The application effectiveness of this teaching platform was tested, and the results showed that the propagation time of the big data platform takes only 112ms, the start-up time is 158ms, and the highest access number ratio can reach 0.82. The passing rate of students using this platform is as high as 73% on average, and the average length of independent learning reaches 6.5 hours. It shows that the online German teaching platform built on the basis of big-number technology has cultivated students’ international vision and cross-cultural communication skills.

In the context of sustainable economic development, while economic globalization brings new vitality to the company, it also makes the company face an increasingly severe external environment. The managers have to shift their focus to capital market investment. The excessive pursuit of investment benefits can easily lead to decision-making errors, resulting in a financial crisis for the company, and even may be forced to delist in severe cases. This article proposes a financial crisis prediction model based on Artificial Bee Colony-recurrent neural network (ABC-RNN) and bidirectional long short-term memory (Bi-LSTM) company with a characteristic attention mechanism. We combined ABC-RNN with Bi-LSTM to extract more temporal feature vectors from financial data. Then we introduced a feature attention mechanism to extract better depth features from financial data; the ABC algorithm is introduced to optimize the weight and bias of RNN to improve the reasoning speed and accuracy. The experiment shows that the prediction accuracy and recall of the model on the test set have reached 88.94% and 88.23%, respectively, which has good prediction ability. The outcome of this research helps the company to prevent and deal with the financial crisis in time and promote the sustainable development of the market economy.

The integration of solar energy with a power system brings great economic and environmental benefits. However, the high penetration of solar power is challenging due to the operation and planning of the existing power system owing to the intermittence and randomicity of solar power generation. Achieving accurate predictions for power generation is important to provide high-quality electric energy for end-users. Therefore, in this paper, we introduce a deep learning-based dual-stream convolutional neural network (CNN) and long short-term nemory (LSTM) network followed by a self-attention mechanism network (DSCLANet). Here, CNN is used to learn spatial patterns and LSTM is incorporated for temporal feature extraction. The output spatial and temporal feature vectors are then fused, followed by a self-attention mechanism to select optimal features for further processing. Finally, fully connected layers are incorporated for short-term solar power prediction. The performance of DSCLANet is evaluated on DKASC Alice Spring solar datasets, and it reduces the error rate up to 0.0136 MSE, 0.0304 MAE, and 0.0458 RMSE compared to recent state-of-the-art methods.

目的 在人体行为识别研究中，利用多模态方法将深度数据与骨骼数据相融合，可有效提高动作的识别率。针对深度图像信息数据量大、冗余度高等问题，提出一种通过获取关键时程信息动作帧序列降低冗余的算法，即质心运动路径松弛算法，并根据不同模态数据的特点，提出一种新的时空特征表示方法。方法 质心运动路径松弛算法根据质心在相邻帧之间的运动距离，计算图像差分后获得的活跃部分的相似系数，然后剔除掉相似度高的帧，获得足以表达行为的关键时程信息。根据图像动态部分的变化特性、人体各部分在运动中的协同性和局部显著性特征构建一种新的时空特征表示方法。结果 在MSR-Action3D数据集上对本文方法的效果进行验证。在3个子集中进行交叉验证的平均分类识别率为95.743 2%，分别比Multi-fused，CovP3DJ，D3D-LSTM（densely connected 3DCNN and long short-term memory），Joint Subset Selection方法高2.443 2%，4.763 2%，0.343 2%，0.213 2%。本文方法在使用完整数据集的扩展实验中进行交叉验证的分类识别率为93.040 3%，具有很好的鲁棒性。结论 实验结果表明，本文提出的去冗余算法在降低冗余后提升了识别效果，提取的特征之间具有相关性低的特点，在组合识别中具有良好的互补性，有效提高了分类识别的精确度。;Objective Human body motion-related recognition has been developing in the context of computer vision and pattern recognition like auxiliary human-computer interaction，motion analysis，intelligent monitoring，and virtual reality. To obtain two-dimensional information for its behavioral recognition，conventional motion behavior recognition is mainly used the RGB image sequence captured by RGB camera. To improve the ability to detect short-duration fragments，current feature descriptors for RGB image sequences are employed to characterize human behavior，such as histogram of oriented gradient（HOG），histogram of optical flow （HOF），and a three-dimensional feature pyramid. Some researchers are focused on the feature that image depth is insensitive to ambient light since RGB images are oriented to behavior image sequences of objects in terms of two-dimensional information. The depth information of the image is coordinated with the features of RGB image to describe the related behavior. Human behavior recognition-relevant multi-modal method can be used to fuse depth data and skeleton data，which can improve the recognition rate of action effectively. Recent depth map is widely used in relevant to human behavior recognition. But，the collection of depth information data is required to be optimized because of time complexity of feature extraction and space complexity of feature storage. To resolve the problems，we develop an algorithm to optimize frames of the depth map and resource consumption. At the same time，a new representation of motion features is facilitated as well according to the motion information of the centroid. Method First，the temporal feature vector is used in terms of depth map sequence-extracted time sequence information. The centroid motion path relaxation algorithm is used to realize depth image de-duplication and de redundancy，and the skeleton map-extracted spatial structure feature vector from are spliced to form the spatio-temporal feature input. Next，spatial features are extracted in terms of the original skeleton points coordinates-spliced three-channel spatial feature map. Finally，the fusion probability of spatio-temporal features and spatial features is used for classification and recognition. Our centroid motion path relaxation algorithm is focused on the optimization of redundant information，the time complexity of feature extraction，and the space complexity of feature storage. For the skeleton data，the global feature of motion direction is proposed to fully reflect the integrity and coordination of limb movements. The extracted features are concatenated to obtain the spatio-temporal feature vector，and they can be fused and enhanced through the original coordinates of skeleton points-built three-channel spatial feature map. Its effectiveness is verified on the MSR-Action3D dataset. Result The experimental setting 1 demonstrate that it is 0. 826 0% higher than the depth motion map（DMM）-local binary pattern（LBP）algorithm，1. 015 2% higher than DMM-CRC（collaborative representation classifier），3. 450 1% higher than gradient local auto correlation（DMM-GLAC） algorithm，0. 605 8% higher than EigenJoint algorithm，and 0. 605 8% higher than space-time auto correlation of gradient （STACOG）algorithm is 10. 624 5% higher. After removing redundancy，the result of experimental setting 1 is 0. 126 1% higher as well. The cross-validation on experimental setting 2 show that the average classification and recognition rate in the three subsets is 95. 743 2%，2. 443 2% higher than multi-fused method，4. 763 2% higher than CovP3DJ method，0. 343 2% higher than D3D-LSTM method，and 0. 213 2% higher than joint subset selection method. For the overall data set，it is 2. 030 3% higher than low latency method，0. 240 3% higher than combination of deep models method，and 2. 340 3% higher than complex network coding method. The experimental setting 2 illustrates that the average classification recognition rate of cross-validation in three subsets is 95. 743 2%，and the classification recognition rate of the complete dataset is 93. 040 3%. Conclusion Our algorithm proposed can improve the recognition effect based on redundancy-optimized，and the featuresextracted have lower correlation mutually，which can improve the accuracy of classification recognition effectively.

Structural damage detection plays an important part in structural health monitoring for engineering structures. However, monitored signals are easily polluted by noise and the damaged data are difficult to obtain. In this work, a novel structural damage detection approach using multisensor spatial–temporal graph-based features and deep graph convolutional networks (DGCNs) is presented. The spatial–temporal graph is constructed by the graph theory based on continuous wavelet transform (CWT) of vibration signals. Then, the multisensor spatial–temporal graph-based feature is extracted based on the Laplacian matrix derived from the spatial–temporal graph of the multisensor data. To overcome the limitation of small data size which obstructed the use of the artificial neural network and convolutional neural network, a DGCN is utilized to classify the damage type of the monitored structure. The extracted multisensor spatial–temporal graph-based feature vector is used to represent the node of the global graph as the input of the DGCN. The node with the same condition of the structure can be classified by using the well-trained DGCN. Experiments of the International Association for Structural Control (IASC)-American Society of Civil Engineers (ASCE) SHM benchmark structure and Qatar steel frame structure in the laboratory are performed to verify the effectiveness of the proposed approach. The experimental results show that the DGCN method can be used to detect structural damage by learning from the constructed global graphs. Comparative experiments demonstrate that the proposed approach performs better than the conventional approach, especially for the limited dataset and noise-polluted case.

Intelligent dual stream CNN and echo state network for anomaly detection

Machine learning based video segmentation of moving scene by motion index using IO detector and shot segmentation

A vehicle motion state prediction algorithm integrating point cloud timing multiview features and multitarget interaction information is proposed in this work to effectively predict the motion states of traffic participants around intelligent vehicles in complex scenes. The algorithm analyzes the characteristics of object motion that are affected by the surrounding environment and the interaction of nearby objects and is based on the complex traffic environment perception dual multiline light detection and ranging (LiDAR) technology. The time sequence aerial view map and time sequence front view depth map are obtained using real-time point cloud information perceived by the LiDAR. Time sequence high-level abstract combination features in the multiview scene are then extracted by an improved VGG19 network model and are fused with the potential spatiotemporal interaction of the multitarget operation state data extraction features detected by the laser radar by using a one-dimensional convolution neural network. A temporal feature vector is constructed as the input data of the bidirectional long-term and short-term memory (BiLSTM) network, and the desired input-output mapping relationship is trained to predict the motion state of traffic participants. According to the test results, the proposed BiLSTM model based on point cloud multiview and vehicle interaction information is better than other methods in predicting the state of target vehicles. The results can provide support for the research to evaluate the risk of intelligent vehicle operation environment.

A hybrid approach for forecasting ship motion using CNN–GRU–AM and GCWOA

Dynamic myocardial perfusion computed tomography (DMP-CT) is an effective medical imaging technique for coronary artery disease diagnosis and therapy guidance. However, the radiation dose received by the patient during repeated CT scans is a widespread concern of radiologists because of the increased risk of cancer. The sparse few-view CT scanning protocol can be a feasible approach to reduce the radiation dose of DMP-CT imaging; however, an advanced reconstruction algorithm is needed. In this paper, a temporal feature prior-aided separated reconstruction method (TFP-SR) for low-dose DMP-CT images reconstruction from sparse few-view sinograms is proposed. To implement the proposed method, the objective perfusion image is divided into the baseline fraction and the enhancement fraction introduced by the arrival of the contrast agent. The core of the proposed TFP-SR method is the utilization of the temporal evolution information that naturally exists in the DMP-CT image sequence to aid the enhancement image reconstruction from limited data. The temporal feature vector of an image pixel is defined by the intensities of this pixel in the pre-reconstructed enhancement sequence, and the connection between two related features is calculated via a zero-mean Gaussian function. A prior matrix is constructed based on the connections between the extracted temporal features and used in the iterative reconstruction of the enhancement images. To evaluate the proposed method, the conventional filtered back-projection algorithm, the total variation regularized PWLS (PWLS-TV) and the prior image constrained compressed sensing are compared in this paper based on studies on a digital extended cardiac-torso (XCAT) thoracic phantom and a preclinical porcine DMP-CT data set that take image misregistration into account. The experimental results demonstrate that the proposed TFP-SR method has superior performance in sparse DMP-CT images reconstruction in terms of image quality and the analyses of the time attenuation curve and hemodynamic parameters.

Electromyography (EMG) signals that obtained by electrodes connected to the forearm are the monitoring of the muscles by the electrical method. These signals are quite useful during the use of prosthesis as a source signal to the moving prosthesis. Therefore, it is essential that classifying the EMG signals with high accuracy by analyzing. This study aims that classifying the individual and combined finger movements using surface EMG signals taken from the surface of the human forearm. EMG signals that belong to 10 different finger movements obtained from eight subjects were used. Firstly, EMG signals have been split into segments by the windowing process, and temporal feature vectors are formed by applying various feature extraction methods to these segments. Feature vectors have been classified with the ensemble bagged tree algorithm, which is a combination of classifiers, to obtain the correct classification decision. As a result of 10-fold cross-validation, with the proposed method, 96.6% overall classification accuracy was achieved. The results obtained show that the ensemble classifier can be used successfully in determining finger movements when compared with similar studies.

Online reviews and comments after product sales have become very important for making buying and selling decisions. Fake reviews will affect such decisions due to deceptive information, leading to financial losses for the consumers. Identification of fake reviews has thus received a great deal of attention in recent years. However, most websites have only focused on dealing with problematic reviews and comments. Amazon and Yelp would only remove possible fake reviews without questioning the sellers who could continue posting deceptive reviews for business purposes. In this paper, we propose a method for the detection of fake reviews based on review records associated with products. We first analyze the characteristics of review data using a crawled Amazon China dataset, which shows that the patterns of review records for products are similar in normal situations. In the proposed method, we first extract the review records of products to a temporal feature vector and then develop an isolation forest algorithm to detect outlier reviews by focusing on the differences between the patterns of product reviews to identify outlier reviews. We will verify the effectiveness of our method and compare it to some existing temporal outlier detection methods using the crawled Amazon China dataset. We will also study the impact caused by the parameter selection of the review records. Our work provides a new perspective of outlier review detection and our experiment demonstrates the effectiveness of our proposed method.

Pulmonary embolism (PE) is a life-threatening clinical problem, and computed tomographic imaging is the standard for diagnosis. Clinical decision support rules based on PE risk-scoring models have been developed to compute pretest probability but are underused and tend to underperform in practice, leading to persistent overuse of CT imaging for PE. To develop a machine learning model to generate a patient-specific risk score for PE by analyzing longitudinal clinical data as clinical decision support for patients referred for CT imaging for PE. In this diagnostic study, the proposed workflow for the machine learning model, the Pulmonary Embolism Result Forecast Model (PERFORM), transforms raw electronic medical record (EMR) data into temporal feature vectors and develops a decision analytical model targeted toward adult patients referred for CT imaging for PE. The model was tested on holdout patient EMR data from 2 large, academic medical practices. A total of 3397 annotated CT imaging examinations for PE from 3214 unique patients seen at Stanford University hospitals and clinics were used for training and validation. The models were externally validated on 240 unique patients seen at Duke University Medical Center. The comparison with clinical scoring systems was done on randomly selected 100 outpatient samples from Stanford University hospitals and clinics and 101 outpatient samples from Duke University Medical Center. Prediction performance of diagnosing acute PE was evaluated using ElasticNet, artificial neural networks, and other machine learning approaches on holdout data sets from both institutions, and performance of models was measured by area under the receiver operating characteristic curve (AUROC). Of the 3214 patients included in the study, 1704 (53.0%) were women from Stanford University hospitals and clinics; mean (SD) age was 60.53 (19.43) years. The 240 patients from Duke University Medical Center used for validation included 132 women (55.0%); mean (SD) age was 70.2 (14.2) years. In the samples for clinical scoring system comparisons, the 100 outpatients from Stanford University hospitals and clinics included 67 women (67.0%); mean (SD) age was 57.74 (19.87) years, and the 101 patients from Duke University Medical Center included 59 women (58.4%); mean (SD) age was 73.06 (15.3) years. The best-performing model achieved an AUROC performance of predicting a positive PE study of 0.90 (95% CI, 0.87-0.91) on intrainstitutional holdout data with an AUROC of 0.71 (95% CI, 0.69-0.72) on an external data set from Duke University Medical Center; superior AUROC performance and cross-institutional generalization of the model of 0.81 (95% CI, 0.77-0.87) and 0.81 (95% CI, 0.73-0.82), respectively, were noted on holdout outpatient populations from both intrainstitutional and extrainstitutional data. The machine learning model, PERFORM, may consider multitudes of applicable patient-specific risk factors and dependencies to arrive at a PE risk prediction that generalizes to new population distributions. This approach might be used as an automated clinical decision-support tool for patients referred for CT PE imaging to improve CT use.

Purpose Previous dynamic prediction models rarely handle multi-period data with different intervals, and the large-scale patient hospital records are not effectively used to improve the prediction performance. This paper aims to focus on the prediction of cardiovascular disease using the improved long short-term memory (LSTM) model. Design/methodology/approach A new model based on the traditional LSTM was proposed to predict cardiovascular disease. The irregular time interval is smoothed to obtain the time parameter vector, and it is used as the input of the forgetting gate of LSTM to overcome the prediction obstacle caused by the irregular time interval. Findings The experimental results show that the dynamic prediction model proposed in this paper obtained a significant better classification performance compared with the traditional LSTM model. Originality/value In this paper, the authors improved the LSTM by smoothing the irregular time between different medical stages of the patient to obtain the temporal feature vector.

A new feature vector based on motion structure of video sequences is proposed which can be used independently or along with motion vectors in video analysis and retrieval applications. The proposed feature vector employs size of prediction units (PUs) in P and B frames of high efficiency video coding (HEVC) coded videos. The feature vector can be extracted by partial decoding of compressed video in the same decoding stage as motion vectors. Video retrieval application is selected to evaluate the performance of the new feature vector. In experimental evaluations, the proposed temporal feature vector independently shows on average 12% improvement in performance against motion vectors and by using it along with motion vectors, the performance of motion vector improves on average by 32% in average normalised modified retrieval rank (ANMRR) measure.