Search

Different from other computer vision tasks, action recognition needs to process larger-scale video data. How to extract and analyze the effective parts from a huge amount of video information is the main difficulty of action recognition technology. In recent years, due to the outstanding performance of Graph Convolutional Networks (GCN) in many fields, a new solution to the action recognition algorithm has emerged. However, in current GCN models, the constant physical adjacency matrix makes it difficult to mine synergistic relationships between key points that are not directly connected in physical space. Additionally, a simple time connection of skeleton data from different frames makes each frame in the video contribute equally to the recognition results, which increases the difficulty of distinguishing action stages. In this paper, the information extraction ability of the model has been optimized in the space domain and time domain, respectively. In the space domain, an Adjacency Matrix Generation (AMG) module, which can pre-analyze node sets and generate an adaptive adjacency matrix, has been proposed. The adaptive adjacency matrix can help the graph convolution model to extract the synergistic information between the key points that are crucial for recognition. In the time domain, the Time Domain Attention (TDA) mechanism has been designed to calculate the time-domain weight vector through double pooling channels and complete the weights of key point sequences. Furthermore, performance of the improved TDA-AMG-GCN modules has been verified on the NTU-RGB+D dataset. Its detection accuracy at the CS and CV divisions reached 84.5% and 89.8%, respectively, with an average level higher than other commonly used detection methods at present. [ABSTRACT FROM AUTHOR]

Graph convolutional networks (GCNs), which extend convolutional neural networks (CNNs) to non-Euclidean structures, have been utilized to promote skeleton-based human action recognition research and have made substantial progress in doing so. However, there are still some challenges in the construction of recognition models based on GCNs. In this paper, we propose an enhanced adjacency matrix-based graph convolutional network with a combinatorial attention mechanism (CA-EAMGCN) for skeleton-based action recognition. Firstly, an enhanced adjacency matrix is constructed to expand the model's perceptive field of global node features. Secondly, a feature selection fusion module (FSFM) is designed to provide an optimal fusion ratio for multiple input features of the model. Finally, a combinatorial attention mechanism is devised. Specifically, our spatial-temporal (ST) attention module and limb attention module (LAM) are integrated into a multi-input branch and a mainstream network of the proposed model, respectively. Extensive experiments on three large-scale datasets, namely the NTU RGB+D 60, NTU RGB+D 120 and UAV-Human datasets, show that the proposed model takes into account both requirements of light weight and recognition accuracy. This demonstrates the effectiveness of our method. [ABSTRACT FROM AUTHOR]

Modeling complex spatial and temporal dependencies in multivariate time series data is crucial for traffic forecasting. Graph convolutional networks have proved to be effective in predicting multivariate time series. Although a predefined graph structure can help the model converge to good results quickly, it also limits the further improvement of the model due to its stationary state. In addition, current methods may not converge on some datasets due to the graph structure of these datasets being difficult to learn. Motivated by this, we propose a novel model named Dynamic Correlation Graph Convolutional Network (DCGCN) in this paper. The model can construct adjacency matrices from input data using a correlation coefficient; thus, dynamic correlation graph convolution is used for capturing spatial dependencies. Meanwhile, gated temporal convolution is used for modeling temporal dependencies. Finally, we performed extensive experiments to evaluate the performance of our proposed method against ten existing well-recognized baseline methods using two original and four public datasets. [ABSTRACT FROM AUTHOR]

Accurate forecasting of taxi demand has facilitated the rational allocation of urban public transport resources, reduced congestion in urban transport networks, and shortened passenger waiting time. However, virtual station discovery and modelling of the demand when forecasting through graph convolutional neural networks remains challenging. In this study, the virtual station discovery problem was addressed by using a two-stage clustering approach, which considers the geographical and load characteristics of taxi demand. Furthermore, a fusion model combining non-negative matrix decomposition and a graph convolutional neural network was proposed in order to extract the features of the nodes for dimension reduction and adaptive adjacency matrix computation. By the construction of a local processing structure, further extraction of the local characteristics of the demand was achieved. The experimental results show that the method in this study outperforms state-of-the-art methods in terms of the root mean square error and average absolute value error. Therefore, the model proposed in this study is able to achieve accurate forecasting of taxi demand. [ABSTRACT FROM AUTHOR]

The research on node importance identification for temporal networks has attracted much attention. In this work, combined with the multi-layer coupled network analysis method, an optimized supra-adjacency matrix (OSAM) modeling method was proposed. In the process of constructing an optimized super adjacency matrix, the intra-layer relationship matrixes were improved by introducing the edge weight. The inter-layer relationship matrixes were formed by improved similarly and the inter-layer relationship is directional by using the characteristics of directed graphs. The model established by the OSAM method accurately expresses the structure of the temporal network and considers the influence of intra- and inter-layer relationships on the importance of nodes. In addition, an index was calculated by the average of the sum of the eigenvector centrality indices for a node in each layer and the node importance sorted list was obtained from this index to express the global importance of nodes in temporal networks. The experimental results on three real temporal network datasets Enron, Emaildept3, and Workspace showed that compared with the SAM and the SSAM methods, the OSAM method has a faster message propagation rate and larger message coverage and better SIR and NDCG@10 indicators. [ABSTRACT FROM AUTHOR]

Traffic prediction is important in applications such as traffic management, route planning, and traffic flow optimization. Traffic speed prediction is an important part of traffic forecasting, which has always been a challenging problem due to the complexity and dynamics of traffic systems. In order to predict traffic speed more accurately, we propose a traffic speed prediction model based on a multi-head attention mechanism and weighted adjacency matrix: MAT-WGCN. MAT-WGCN first uses GCN to extract the road spatial features in the weighted adjacency matrix, and it uses GRU to extract the correlation between speed and time from the original features. Then, the spatial features extracted by GCN and the temporal features extracted by GRU are fused, and a multi-head attention mechanism is introduced to integrate spatiotemporal features, collect and summarize spatiotemporal road information, and realize traffic speed prediction. In this study, the prediction performance of MAT-WGCN was tested on two real datasets, EXPY-TKY and METR-LA, and compared with the performance of traditional methods such as HA and SVR that do not combine spatial features, as well as T-GCN, A3T-GCN, and newer methods such as GCN and NA-DGRU that combine spatial features. The experimental results demonstrate that MAT-WGCN can capture the temporal and spatial characteristics of road conditions, thus enabling accurate traffic speed predictions. Furthermore, the incorporation of a multi-head attention mechanism significantly enhances the robustness of our model. [ABSTRACT FROM AUTHOR]

Background: Post-translational modification (PTM) is a biological process that is associated with the modification of proteome, which results in the alteration of normal cell biology and pathogenesis. There have been numerous PTM reports in recent years, out of which, lysine phosphoglycerylation has emerged as one of the recent developments. The traditional methods of identifying phosphoglycerylated residues, which are experimental procedures such as mass spectrometry, have shown to be time-consuming and cost-inefficient, despite the abundance of proteins being sequenced in this post-genomic era. Due to these drawbacks, computational techniques are being sought to establish an effective identification system of phosphoglycerylated lysine residues. The development of a predictor for phosphoglycerylation prediction is not a first, but it is necessary as the latest predictor falls short in adequately detecting phosphoglycerylated and non-phosphoglycerylated lysine residues., Results: In this work, we introduce a new predictor named RAM-PGK, which uses sequence-based information relating to amino acid residues to predict phosphoglycerylated and non-phosphoglycerylated sites. A benchmark dataset was employed for this purpose, which contained experimentally identified phosphoglycerylated and non-phosphoglycerylated lysine residues. From the dataset, we extracted the residue adjacency matrix pertaining to each lysine residue in the protein sequences and converted them into feature vectors, which is used to build the phosphoglycerylation predictor., Conclusion: RAM-PGK, which is based on sequential features and support vector machine classifiers, has shown a noteworthy improvement in terms of performance in comparison to some of the recent prediction methods. The performance metrics of the RAM-PGK predictor are: 0.5741 sensitivity, 0.6436 specificity, 0.0531 precision, 0.6414 accuracy, and 0.0824 Mathews correlation coefficient.

Adjacency matrix visualization is a common method for presenting graph data, and the Focus+Context technique can be used to explore the details of the ROI (region of interest). Embedded views and multi-view approaches are usually applied when locating and comparing attributes among multiple nodes. However, the embedded view has an issue of edge occlusion, while the multi-view would cause repeated perspective switching. In this paper, we propose a Multivariate Fence (MVF) model as a focus view of the adjacency matrix to locate and compare attributes among nodes. An additional spatial parallel coordinate is added to the 2D adjacency matrix in an immersive environment so that the attribute information can be shown in a single view without blocking edge information. We also conduct a user study to evaluate the performance of the MVF. The results show that the MVF has better efficiency and accuracy in locating and comparing the multivariate adjacency matrix in the immersive environment against the existing focus model. Moreover, the MVF model is easier to understand and is preferred by users. [ABSTRACT FROM AUTHOR]

The maximum eigenvalue of the adjacency matrix (AM) has been supposed to contain rich information about the corresponding network. An experimental study focused on revealing the meaning and application of the maximum eigenvalue is missing. To this end, AM was constructed using mutual information (MI) to determine the functional connectivity with electroencephalogram (EEG) data recorded with a mental fatigue model, and then was converted into both binary and weighted brain functional network (BFN) and corresponding random networks (RNs). Both maximum eigenvalue and corresponding network characters in BFNs and RNs were considered to explore the changes during the formation of mental fatigue. The results indicated that large maximum eigenvalue means more edges in the corresponding network, along with a high degree and a short characteristic path length both in weighted and binary BFNs. Interestingly, the maximum eigenvalue of AM was always a little larger than that of the corresponding random matrix (RM), and had an obvious linearity with the sum of the AM elements, indicating that the maximum eigenvalue can be able to distinguish the network structures which have the same mean degree. What is more, the maximum eigenvalue, which increased with the deepening of mental fatigue, can become a good indicator for mental fatigue estimation., Competing Interests: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Accurately predicting the active power output of offshore wind power is of great significance for reducing the uncertainty in new power systems. By utilizing the spatiotemporal correlation characteristics among wind turbine unit outputs, this paper embeds the Diffusion Convolutional Neural Network (DCNN) into the Gated Recurrent Unit (GRU) for the feature extraction of spatiotemporal correlations in wind turbine unit outputs. It also combines graph structure learning to propose a sequence-to-sequence model for ultra-short-term power prediction in large offshore wind farms. Firstly, the electrical connection graph within the wind farm is used to preliminarily determine the reference adjacency matrix for the wind turbine units within the farm, injecting prior knowledge of the adjacency matrix into the model. Secondly, a convolutional neural network is utilized to convolve the historical curves of units within the farm along the time dimension, outputting a unit connection probability vector. The Gumbel–softmax reparameterization method is then used to make the probability vector differentiable, thereby generating an optimal adjacency matrix for the prediction task based on the probability vector. At the same time, the difference between the two adjacency matrices is added as a regularization term to the loss function to reduce model overfitting. The simulation of actual cases shows that the proposed model has good predictive performance in ultra-short-term power prediction for large offshore wind farms. [ABSTRACT FROM AUTHOR]

In this article, we discuss the spectral properties of the general extended adjacency matrix for chain graphs. In particular, we discuss the eigenvalues of the general extended adjacency matrix of the chain graphs and obtain its general extended adjacency inertia. We obtain bounds for the largest and the smallest general extended adjacency eigenvalues and characterize the extremal graphs. We also obtain a lower bound for the spread of the general extended adjacency matrix. We characterize chain graphs with all the general extended adjacency eigenvalues being simple and chain graphs that are non-singular under the general extended adjacency matrix. Further, we determine the explicit formula for the determinant and the trace of the square of the general extended adjacency matrix of chain graphs. Finally, we discuss the energy of the general extended adjacency matrix and obtain some bounds for it. We characterize the extremal chain graphs attaining these bounds. [ABSTRACT FROM AUTHOR]

In addressing the challenges of trajectory prediction in multi-ship interaction scenarios and aiming to improve the accuracy of multi-ship trajectory prediction, this paper proposes a multi-ship trajectory prediction model, GL-STGCNN. The GL-STGCNN model employs a ship interaction adjacency matrix extraction module to obtain a more reasonable ship interaction adjacency matrix. Additionally, after obtaining the distribution of predicted trajectories using the model, a model predictive control trajectory correction method is introduced to enhance the accuracy and reasonability of the predicted trajectories. Through quantitative analysis of different datasets, it was observed that GL-STGCNN outperforms previous prediction models with a 31.8% improvement in the average displacement error metric and a 16.8% improvement in the final displacement error metric. Furthermore, trajectory correction through model predictive control shows a performance boost of 44.5% based on the initial predicted trajectory distribution. While GL-STGCNN excels in multi-ship interaction trajectory prediction by reasonably modeling ship interaction adjacency matrices and employing trajectory correction, its performance may vary in different datasets and ship motion patterns. Future work could focus on adapting the model's ship interaction adjacency matrix modeling to diverse environmental scenarios for enhanced performance. [ABSTRACT FROM AUTHOR]

In this paper, we present an innovative transmit power control scheme, based on optimization theory, for wireless sensor networks (WSNs) which use carrier sense multiple access (CSMA) with collision avoidance (CA) as medium access control (MAC) protocol. In particular, we focus on schemes where several remote nodes send data directly to a common access point (AP). Under the assumption of finite overall network transmit power and low traffic load, we derive the optimal transmit power allocation strategy that minimizes the packet error rate (PER) at the AP. This approach is based on modeling the CSMA/CA MAC protocol through a finite state machine and takes into account the network adjacency matrix, depending on the transmit power distribution and determining the network connectivity. It will be then shown that the transmit power allocation problem reduces to a convex constrained minimization problem. Our results show that, under the assumption of low traffic load, the power allocation strategy, which guarantees minimal delay, requires the maximization of network connectivity, which can be equivalently interpreted as the maximization of the number of non-zero entries of the adjacency matrix. The obtained theoretical results are confirmed by simulations for unslotted Zigbee WSNs. [ABSTRACT FROM AUTHOR]

The purpose of this study is to develop an evaluation method for assessing the relative relationship between gear tooth shape deviations on every gear tooth using network theory. Our previous study introduced a method for representing the phase difference between each helix deviation as a network, demonstrating that it is possible to identify the relative relationship between gear tooth deviations. However, there has been no in-depth analysis of the impact of injection molding on the gear's phase difference network. In this paper, we begin by measuring the gear tooth shape deviation, calculating the correlation coefficient, and expressing it as an adjacency matrix. When the adjacency matrix was visualized and displayed as a pixel plot, a periodic pattern was formed. The relationship between the position of the gate used to inject molten resin material into the injection mold and the pixel plot were then investigated in detail, and it was confirmed that the helix deviation network of a gear manufactured with the injection molding process is useful as a new indicator for the manufacturing error of injection-molded plastic gear. [ABSTRACT FROM AUTHOR]

In the paper, we introduce the signless Laplacian A B C -matrix Q ̃ (G) = D ¯ (G) + A ̃ (G) , where D ¯ (G) is the diagonal matrix of A B C -degrees and A ̃ (G) is the A B C -matrix of G. The eigenvalues of the matrix Q ̃ (G) are the signless Laplacian A B C -eigenvalues of G. We give some basic properties of the matrix Q ̃ (G) , which includes relating independence number and clique number with signless Laplacian A B C -eigenvalues. For bipartite graphs, we show that the signless Laplacian A B C -spectrum and the Laplacian A B C -spectrum are the same. We characterize the graphs with exactly two distinct signless Laplacian A B C -eigenvalues. Also, we consider the problem of the characterization of the graphs with exactly three distinct signless Laplacian A B C -eigenvalues and solve it for bipartite graphs and, in some cases, for non-bipartite graphs. We also introduce the concept of the trace norm of the matrix Q ̃ (G) − t r (Q ̃ (G)) n I , called the signless Laplacian A B C -energy of G. We obtain some upper and lower bounds for signless Laplacian A B C -energy and characterize the extremal graphs attaining it. Further, for graphs of order at most 6, we compare the signless Laplacian energy and the A B C -energy with the signless Laplacian A B C -energy and found that the latter behaves well, as there is a single pair of graphs with the same signless Laplacian A B C -energy unlike the 26 pairs of graphs with same signless Laplacian energy and eight pairs of graphs with the same A B C -energy. [ABSTRACT FROM AUTHOR]

Given that fingerprint localization methods can be effectively modeled as supervised learning problems, machine learning has been employed for indoor localization tasks based on fingerprint methods. However, it is often challenging for popular machine learning models to effectively capture the unstructured data features inherent in fingerprint data that are generated in diverse propagation environments. In this paper, we propose an indoor localization algorithm based on a high-order graph neural network (HoGNNLoc) to enhance the accuracy of indoor localization and improve localization stability in dynamic environments. The algorithm first designs an adjacency matrix based on the spatial relative locations of access points (APs) to obtain a graph structure; on this basis, a high-order graph neural network is constructed to extract and aggregate the features; finally, the designed fully connected network is used to achieve the regression prediction of the location of the target to be located. The experimental results on our self-built dataset show that the proposed algorithm achieves localization accuracy within 1.29 m at 80% of the cumulative distribution function (CDF) points. The improvements are 59.2%, 51.3%, 36.1%, and 22.7% compared to the K-nearest neighbors (KNN), deep neural network (DNN), simple graph convolutional network (SGC), and graph attention network (GAT). Moreover, even with a 30% reduction in fingerprint data, the proposed algorithm exhibits stable localization performance. On a public dataset, our proposed localization algorithm can also show better performance. [ABSTRACT FROM AUTHOR]

Let G be a graph with the adjacency matrix A (G) , and let D (G) be the diagonal matrix of the degrees of G. Nikiforov first defined the matrix A α (G) as A α (G) = α D (G) + (1 − α) A (G) , 0 ≤ α ≤ 1 , which shed new light on A (G) and Q (G) = D (G) + A (G) , and yielded some surprises. The α − adjacency energy E A α (G) of G is a new invariant that is calculated from the eigenvalues of A α (G) . In this work, by combining matrix theory and the graph structure properties, we provide some upper and lower bounds for E A α (G) in terms of graph parameters (the order n, the edge size m, etc.) and characterize the corresponding extremal graphs. In addition, we obtain some relations between E A α (G) and other energies such as the energy E (G) . Some results can be applied to appropriately estimate the α -adjacency energy using some given graph parameters rather than by performing some tedious calculations. [ABSTRACT FROM AUTHOR]

For a simple finite graph G, the generalized adjacency matrix is defined as A α (G) = α D (G) + (1 − α) A (G) , α ∈ [ 0 , 1 ] , where A (G) and D (G) are respectively the adjacency matrix and diagonal matrix of the vertex degrees. The A α -spread of a graph G is defined as the difference between the largest eigenvalue and the smallest eigenvalue of the A α (G) . In this paper, we answer the question posed in (Lin, Z.; Miao, L.; Guo, S. Bounds on the A α -spread of a graph. Electron. J. Linear Algebra2020, 36, 214–227). Furthermore, we show that the path graph, P n , has the smallest S (A α) among all trees of order n. We establish a relationship between S (A α) and S (A). We obtain several bounds for S (A α) . [ABSTRACT FROM AUTHOR]

Emotion is one of the most important higher cognitive functions of the human brain and plays an important role in transaction processing and decisions. In traditional emotion recognition studies, the frequency band features in EEG signals have been shown to have a high correlation with emotion production. However, traditional emotion recognition methods cannot satisfactorily solve the problem of individual differences in subjects and data heterogeneity in EEG, and subject-independent emotion recognition based on EEG signals has attracted extensive attention from researchers. In this paper, we propose a subject-independent emotion recognition model based on adaptive extraction of layer structure based on frequency bands (BFE-Net), which is adaptive in extracting EEG map features through the multi-graphic layer construction module to obtain a frequency band-based multi-graphic layer emotion representation. To evaluate the performance of the model in subject-independent emotion recognition studies, extensive experiments are conducted on two public datasets including SEED and SEED-IV. The experimental results show that in most experimental settings, our model has a more advanced performance than the existing studies of the same type. In addition, the visualization of brain connectivity patterns reveals that some of the findings are consistent with previous neuroscientific validations, further validating the model in subject-independent emotion recognition studies. [ABSTRACT FROM AUTHOR]

This report describes a new set of macromolecular descriptors of relevance to protein QSAR/QSPR studies, protein's quadratic indices. These descriptors are calculated from the macromolecular pseudograph's alpha-carbon atom adjacency matrix. A study of the protein stability effects for a complete set of alanine substitutions in Arc repressor illustrates this approach. Quantitative Structure-Stability Relationship (QSSR) models allow discriminating between near wild-type stability and reduced-stability A-mutants. A linear discriminant function gives rise to excellent discrimination between 85.4% (35/41)and 91.67% (11/12) of near wild-type stability/reduced stability mutants in training and test series, respectively. The model's overall predictability oscillates from 80.49 until 82.93, when n varies from 2 to 10 in leave-n-out cross validation procedures. This value stabilizes around 80.49% when n was > 6. Additionally, canonical regression analysis corroborates the statistical quality of the classification model (Rcanc = 0.72, p-level <0.0001). This analysis was also used to compute biological stability canonical scores for each Arc A-mutant. On the other hand, nonlinear piecewise regression model compares favorably with respect to linear regression one on predicting the melting temperature (tm)of the Arc A-mutants. The linear model explains almost 72% of the variance of the experimental tm (R = 0.85 and s = 5.64) and LOO press statistics evidenced its predictive ability (q2 = 0.55 and scv = 6.24). However, this linear regression model falls to resolve t(m) predictions of Arc A-mutants in external prediction series. Therefore, the use of nonlinear piecewise models was required. The tm values of A-mutants in training (R = 0.94) and test(R = 0.91) sets are calculated by piecewise model with a high degree of precision. A break-point value of 51.32 degrees C characterizes two mutants' clusters and coincides perfectly with the experimental scale. For this reason, we can use the linear discriminant analysis and piecewise models in combination to classify and predict the stability of the mutants' Arc homodimers. These models also permit the interpretation of the driving forces of such a folding process. The models include protein's quadratic indices accounting for hydrophobic (z1), bulk-steric (z2), and electronic (z3) features of the studied molecules. Preponderance of z1 and z3 over z2 indicates the higher importance of the hydrophobic and electronic side chain terms in the folding of the Arc dimer. In this sense, developed equations involve short-reaching (k < or = 3), middle- reaching (3 < k < or = 7) and far-reaching (k= 8 or greater) z1, 2, 3-protein's quadratic indices. This situation points to topologic/topographic protein's backbone interactions control of the stability profile of wild-type Arc and its A-mutants. Consequently, the present approach represents a novel and very promising way to mathematical research in biology sciences.

In this paper we describe the application in QSPR/QSAR studies of a new group of molecular descriptors: atom, atom-type and total linear indices of the molecular pseudograph's atom adjacency matrix. These novel molecular descriptors were used for the prediction of boiling point and partition coefficient (log P), specific rate constant (log k), and antibacterial activity of 28 alkyl-alcohols and 34 derivatives of 2-furylethylenes,respectively. For this purpose two quantitative models were obtained to describe the alkyl-alcohols' boiling points. The first one includes only two total linear indices and showed a good behavior from a statistical point of view (R(2) = 0.984, s = 3.78, F = 748.57,q(2) = 0.981, and s(cv) = 3.91). The second one includes four variables [3 global and 1 local(heteroatom) linear indices] and it showed an improvement in the description of physical property (R(2) = 0.9934, s = 2.48, F = 871.96, q(2) = 0.990, and s(cv) = 2.79). Later, linear multiple regression analysis was also used to describe log P and log k of the 2-furyl-ethylenes derivatives. These models were statistically significant [(R(2) = 0.984, s = 0.143, and F = 113.38) and (R(2) = 0.973, s = 0.26 and F = 161.22), respectively] and showed very good stability to data variation in leave-one-out (LOO) cross-validation experiment [(q(2) = 0.93.8 and scv = 0.178) and (q(2) = 0.948 and s(cv) = 0.33), respectively]. Finally, a linear discriminant model for classifying antibacterial activity of these compounds was also achieved with the use of the atom and atom-type linear indices. The global percent of good classification in training and external test set obtained was of 94.12% and 100.0%, respectively. The comparison with other approaches (connectivity indices, total and local spectral moments, quantum chemical descriptors, topographic indices and E- state/biomolecular encounter parameters) reveals a good behavior of our method. The approach described in this paper appears to be a very promising structural invariant, useful for QSPR/QSAR studies and computer-aided "rational" drug design.

Extracting water bodies from synthetic aperture radar (SAR) images plays a crucial role in the management of water resources, flood monitoring, and other applications. Recently, transformer-based models have been extensively utilized in the remote sensing domain. However, due to regular patch-partition and weak inductive bias, transformer-based models face challenges such as edge serration and high data dependency when used for water body extraction from SAR images. To address these challenges, we introduce a new model, the Superpixel-based Transformer (SPT), based on the adaptive characteristic of superpixels and knowledge constraints of the adjacency matrix. (1) To mitigate edge serration, the SPT replaces regular patch partition with superpixel segmentation to fully utilize the internal homogeneity of superpixels. (2) To reduce data dependency, the SPT incorporates a normalized adjacency matrix between superpixels into the Multi-Layer Perceptron (MLP) to impose knowledge constraints. (3) Additionally, to integrate superpixel-level learning from the SPT with pixel-level learning from the CNN, we combine these two deep networks to form SPT-UNet for water body extraction. The results show that our SPT-UNet is competitive compared with other state-of-the-art extraction models, both in terms of quantitative metrics and visual effects. [ABSTRACT FROM AUTHOR]

The main functions of the space information system, such as providing the backbone transmission, broadband access, and global connectivity, are realized based on the network topology. Thus, it is necessary to recognize the temporal dynamics of the network topology. A temporal continuity expression method is proposed to describe the topological dynamic characteristics of the network in space information systems. Based on orbit dynamics, a time-dependent adjacency matrix of the space information system can be established by introducing the geometric linkable factor, the link distance intensity factor, and the relative angular velocity factor of the node. The adjacency matrix describes the dynamic characteristics from two layers: one is the physical layer using a time-dependent function, which represents the feasibility of inter-satellite link construction in the system cycle; the other one is the transport layer, described by a piecewise continuous function that varies with time, which characterizes the link quality during the connection period between two satellites. The results show that compared with the existing network topology description methods, the proposed method describes the network topology more accurately, which can distinguish the network topology characteristics at any time, and is more conducive to the understanding and application of the network topology of the space information system. [ABSTRACT FROM AUTHOR]

The Graph Convolutional Neural Networks (GCN) method has shown excellent performance in the field of deep learning, and using graphs to represent speech data is a computationally efficient and scalable approach. In order to enhance the adequacy of graph neural networks in extracting speech emotional features, this paper proposes a Temporal-Spatial Learnable Graph Convolutional Neural Network (TLGCNN) for speech emotion recognition. TLGCNN firstly utilizes the Open-SMILE toolkit to extract frame-level speech emotion features. Then, a bidirectional long short-term memory (Bi LSTM) network is used to process the long-term dependencies of speech features which can further extract deep frame-level emotion features. The extracted frame-level emotion features are then input into subsequent network through two pathways. Finally, one pathway constructs the extracted frame-level deep emotion feature vectors into a graph structure applying an adaptive adjacency matrix to catch latent spatial connections, while the other pathway concatenates emotion feature vectors with graph-level embedding obtained from learnable graph convolutional neural network for prediction and classification. Through these two pathways, TLGCNN can simultaneously obtain temporal speech emotional information through Bi-LSTM and spatial speech emotional information through Learnable Graph Convolutional Neural (LGCN) network. Experimental results demonstrate that this method achieves weighted accuracy of 66.82% and 58.35% on the IEMOCAP and MSP-IMPROV databases, respectively. [ABSTRACT FROM AUTHOR]

Hydropower units are the core equipment of hydropower stations, and research on the fault prediction and health management of these units can help improve their safety, stability, and the level of reliable operation and can effectively reduce costs. Therefore, it is necessary to predict the swing trend of these units. Firstly, this study considers the influence of various factors, such as electrical, mechanical, and hydraulic swing factors, on the swing signal of the main guide bearing y-axis. Before swing trend prediction, the multi-index feature selection algorithm is used to obtain suitable state variables, and the low-dimensional effective feature subset is obtained using the Pearson correlation coefficient and distance correlation coefficient algorithms. Secondly, the dilated convolution graph neural network (DCGNN) algorithm, with a dilated convolution graph, is used to predict the swing trend of the main guide bearing. Existing GNN methods rely heavily on predefined graph structures for prediction. The DCGNN algorithm can solve the problem of spatial dependence between variables without defining the graph structure and provides the adjacency matrix of the graph learning layer simulation, avoiding the over-smoothing problem often seen in graph convolutional networks; furthermore, it effectively improves the prediction accuracy. The experimental results showed that, compared with the RNN-GRU, LSTNet, and TAP-LSTM algorithms, the MAEs of the DCGNN algorithm decreased by 6.05%, 6.32%, and 3.04%; the RMSEs decreased by 9.21%, 9.01%, and 2.83%; and the CORR values increased by 0.63%, 1.05%, and 0.37%, respectively. Thus, the prediction accuracy was effectively improved. [ABSTRACT FROM AUTHOR]

Electroencephalography (EEG) is recorded by electrodes from different areas of the brain and is commonly used to measure neuronal activity. EEG-based methods have been widely used for emotion recognition recently. However, most current methods for EEG-based emotion recognition do not fully exploit the relationship of EEG channels, which affects the precision of emotion recognition. To address the issue, in this paper, we propose a novel method for EEG-based emotion recognition called CR-GCN: Channel-Relationships-based Graph Convolutional Network. Specifically, topological structure of EEG channels is distance-based and tends to capture local relationships, and brain functional connectivity tends to capture global relationships among EEG channels. Therefore, in this paper, we construct EEG channel relationships using an adjacency matrix in graph convolutional network where the adjacency matrix captures both local and global relationships among different EEG channels. Extensive experiments demonstrate that CR-GCN method significantly outperforms the state-of-the-art methods. In subject-dependent experiments, the average classification accuracies of 94.69% and 93.95% are achieved for valence and arousal. In subject-independent experiments, the average classification accuracies of 94.78% and 93.46% are obtained for valence and arousal. [ABSTRACT FROM AUTHOR]

The demand for accurate and reliable unsupervised image segmentation methods is high. Regardless of whether we are faced with a problem for which we do not have a usable training dataset, or whether it is not possible to obtain one, we still need to be able to extract the desired information from images. In such cases, we are usually gently pushed towards the best possible clustering method, as it is often more robust than simple traditional image processing methods. We investigate the usefulness of combining two clustering methods for unsupervised image segmentation. We use the mixture models to extract the color and spatial image features based on the obtained output segments. Then we construct a similarity matrix (adjacency matrix) based on these features to perform spectral clustering. In between, we propose a label noise correction using Markov random fields. We investigate the usefulness of our method on many hand-crafted images of different objects with different shapes, colorization, and noise. Compared to other clustering methods, our proposal performs better, with 10% higher accuracy. Compared to state-of-the-art supervised image segmentation methods based on deep convolutional neural networks, our proposal proves to be competitive. [ABSTRACT FROM AUTHOR]

We introduce a generalized adjacency matrix in order to distinguish cospectral graphs. Our reasoning is motivated by the work of Johnson and Newman and properties of p-adic numbers. Using a polynomial time algorithm, we comment on computer experiments with which we can distinguish cospectral (non-isomorphic) graphs. [ABSTRACT FROM AUTHOR]

A singular graph G, defined when its adjacency matrix is singular, has important applications in mathematics, natural sciences and engineering. The chemical importance of singular graphs lies in the fact that if the molecular graph is singular, the nullity (the number of the zero eigenvalue) is greater than 0, then the corresponding chemical compound is highly reactive or unstable. By this reasoning, chemists have a great interest in this problem. Thus, the problem of characterization singular graphs was proposed and raised extensive studies on this challenging problem thereafter. The graph obtained by conglutinating the starting vertices of three paths P s 1 , P s 2 , P s 3 into a vertex, and three end vertices into a vertex on the cycle C a 1 , C a 2 , C a 3 , respectively, is denoted as γ (a 1 , a 2 , a 3 , s 1 , s 2 , s 3) . Note that δ (a 1 , a 2 , a 3 , s 1 , s 2) = γ (a 1 , a 2 , a 3 , s 1 , 1 , s 2) , ζ (a 1 , a 2 , a 3 , s) = γ (a 1 , a 2 , a 3 , 1 , 1 , s) , φ (a 1 , a 2 , a 3) = γ (a 1 , a 2 , a 3 , 1 , 1 , 1) . In this paper, we give the necessity and sufficiency that the γ − graph, δ − graph, ζ − graph and φ − graph are singular and prove that the probability that a randomly given γ − graph, δ − graph, ζ − graph or φ − graph being singular is equal to 325 512 , 165 256 , 43 64 , 21 32 , respectively. From our main results, we can conclude that such a γ − graph( δ − graph, ζ − graph, φ − graph) is singular if at least one cycle is a multiple of 4 in length, and surprisingly, the theoretical probability of these graphs being singular is more than half. This result promotes the understanding of a singular graph and may be promising to propel the solutions to relevant application problems. [ABSTRACT FROM AUTHOR]

A Modular Reconfigurable Robot (MRR) composed of standard joint and link modules has the advantages of rapid changeover of assembly configurations for a diversity of application requirements. To tackle the kinematics analysis issues for an MRR, it is essential to develop a generic method to mathematically represent all possible assembly configurations and automatically generate their kinematic models. In this paper, two types of robot modules, i.e., revolute joint modules and rigid link modules, are considered as the basic building blocks of an MRR. The topological structure of an MRR is represented by a graph, termed an Assembly Graph (AG), in which all robot modules, regardless of their types, are treated as vertices, while the connecting interfaces between any two robot modules are treated as edges. Based on graph theory, a modified adjacency matrix, termed an Assembly Adjacency Matrix (AAM), is proposed to represent the assembly configuration of an MRR, in which the three-dimensional assembly information is specified with connecting port vectors for the adjacent modules. A path matrix derived from the AAM of an MRR is employed to describe the connecting sequence of its constituting modules. The local frame representation of the Product-of-Exponentials (POE) formula is employed, which is configuration-independent. Therefore, for an MRR configuration represented with an AAM, its kinematic model will be automatically generated according to its path matrix. The proposed assembly configuration representation and kinematic modeling method are validated through a dual-branch MRR configuration. [ABSTRACT FROM AUTHOR]

We presented a method based on multigraphs to mathematically define a distribution function in time for the generation of data exchange in a special-purpose communication network. This is needed for the modeling and design of communication networks (CNs) consisting of integrated telecommunications and computer networks (ITCN). Simulation models require a precise definition of network traffic communication. An additional problem for describing the network traffic in simulation models is the mathematical model of data distribution, according to which the generation and exchange of certain types and quantities of data are realized. The application of multigraphs enabled the time and quantity of the data distribution to be displayed as operational procedures for a special-purpose communication unit. A multigraph was formed for each data-exchange time and allowed its associated adjacency matrix to be defined. Using the matrix estimation method allowed the mathematical definition of the distribution function values. The application of the described method for the use of multigraphs enabled a more accurate mathematical description of real traffic in communication networks. [ABSTRACT FROM AUTHOR]

The article introduces a new method called the Unique Brain Network Identification Number (UBNIN) for storing binary neuroimaging connectivity data. It compares the UBNIN method with three other methods for storing brain-state data and discusses the pros and cons of each approach. The UBNIN method has three variations with different levels of compactness and information loss. The article concludes that while the UBNIN method has advantages, it requires more storage space compared to other methods and does not provide detailed information about connections between brain areas. The article also includes two tables comparing different properties of brain-state representations. The authors of the paper were involved in various aspects of the research and declare no conflicts of interest. [Extracted from the article]

Entity alignment is an important task in knowledge fusion, which aims to link entities that have the same real-world identity in two knowledge graphs. However, in the process of constructing a knowledge graph, some noise may inevitably be introduced, which must affect the results of the entity alignment tasks. The triple confidence calculation can quantify the correctness of the triples to reduce the impact of the noise on entity alignment. Therefore, we designed a method to calculate the confidence of the triples and applied it to the knowledge representation learning phase of entity alignment. The method calculates the triple confidence based on the pairing rates of the three angles between the entities and relations. Specifically, the method uses the pairing rates of the three angles as features, which are then fed into a feedforward neural network for training to obtain the triple confidence. Moreover, we introduced the triple confidence into the knowledge representation learning methods to improve their performance in entity alignment. For the graph neural network-based method GCN, we considered entity confidence when calculating the adjacency matrix, and for the translation-based method TransE, we proposed a strategy to dynamically adjust the margin value in the loss function based on confidence. These two methods were then applied to the entity alignment, and the experimental results demonstrate that compared with the knowledge representation learning methods without integrating confidence, the confidence-based knowledge representation learning methods achieved excellent performance in the entity alignment task. [ABSTRACT FROM AUTHOR]

Radar emitter signal sorting (RESS) is a crucial process in contemporary electronic battlefield situation awareness. Separating pulses belonging to the same radar emitter from interleaved radar pulse sequences with a lack of prior information, high density, strong overlap, and wide parameter distribution has attracted increasing attention. In order to improve the accuracy of RESS under scenarios with limited labeled samples, this paper proposes an RESS model called TR-RAGCN-AFF-RESS. This model transforms the RESS problem into a pulse-by-pulse classification task. Firstly, a novel weighted adjacency matrix construction method was proposed to characterize the structural relationships between pulse attribute parameters more accurately. Building upon this foundation, two networks were developed: a Transformer(TR)-based interleaved pulse sequence temporal feature extraction network and a residual attention graph convolutional network (RAGCN) for extracting the structural relationship features of attribute parameters. Finally, the attention feature fusion (AFF) algorithm was introduced to fully integrate the temporal features and attribute parameter structure relationship features, enhancing the richness of feature representation for the original pulses and achieving more accurate sorting results. Compared to existing deep learning-based RESS algorithms, this method does not require many labeled samples for training, making it better suited for scenarios with limited labeled sample availability. Experimental results and analysis confirm that even with only 10% of the training samples, this method achieves a sorting accuracy exceeding 93.91%, demonstrating high robustness against measurement errors, lost pulses, and spurious pulses in non-ideal conditions. [ABSTRACT FROM AUTHOR]

This work surveys the use of directed weighted graphs in conducting comparative static analyses. The paper discusses the implementation of a computer-aided process for building spreadsheet-based graph models for inter- and intra-regional financial flows. The graph-theoretic techniques are programmed to enable the interactive visualization and analysis of financial data using Wolfram technologies (i.e., Mathematica software v. 11.3 or later, Wolfram player v. 12 or later). The paper describes the workflow for several interactive visualizations applicable to financial networks. The author provides four programs, written in the Wolfram language, that customize input–output financial models by combining the Manipulate command with built-in Mathematica functions and functions of the IGraph package (IGraph/M). The study is a tutorial article for the generation of a suite of visual schemes that provide patterns, practices, and roadmaps of the financial markets across the globe. [ABSTRACT FROM AUTHOR]

Realizing the distributed adaptive network construction of multi-UAV networks is an urgent challenge, as they lack a reliable common control channel and can only maintain a limited sensing range in crowded electromagnetic environments. Multi-unmanned aerial vehicle (UAV) networks are gaining popularity in many fields. In order to address these issues, this paper proposes a multi-UAV network channel rendezvous algorithm based on average consistency. The goal of the algorithm is to adjust the communication channels of each UAV to converge on the same channel, since the communication link of the multi-UAV network is broken due to interference. The proposed memory-based average consistency (MAC) algorithm utilizes the network adjacency matrix as prior information. Furthermore, for the case where the adjacency matrix is unknown, this paper also proposes the Multi-Radio Average Consensus (MRAC) algorithm, which achieves a beneficial trade-off between rendezvous performance and hardware cost. Simulation results demonstrate that the proposed MAC and MRAC algorithms provide superior network convergence time and scalability in networks of different densities. Finally, a hardware simulation platform based on a multi-UAV network was designed using a software-defined radio platform, and experimental simulations were performed to prove the effectiveness of the MAC algorithm in a real environment. [ABSTRACT FROM AUTHOR]

In road networks, attribute information carried by road segment nodes, such as weather and points of interest (POI), exhibit strong heterogeneity and often involve one-to-many or many-to-one relationships. However, research on such heterogeneity in traffic prediction is relatively limited. Our research examines how varying the network propagation pattern based on the degree of node-to-node heterogeneity of information affects the model prediction performance. Specifically, at the node level, we use knowledge embedding to generate knowledge vectors that quantify the heterogeneity among the attribute information of a node. At the road network level, we calculate a homogeneity adjacency matrix that captures both the topological structure of the road network and the similarity of node heterogeneity. This adjacency matrix assigns different weights to neighbors based on their homogeneity, guiding the propagation of graph convolutional networks (GCN). Finally, we separate the representation of propagation into self-representation and neighbor representation to extract multi-attribute information, including self, homogeneity, and heterogeneity. Experiments on real datasets demonstrate that the incorporation of our homogeneity adjacency matrix leads to a significant improvement in the accuracy of short-term and long-term prediction compared with previous work on homogeneous and single-dimensional information. Furthermore, our approach maintains its performance advantage over baseline models under different embedding dimensions and parameter settings. [ABSTRACT FROM AUTHOR]

Obtaining accurate road conditions is crucial for traffic management, dynamic route planning, and intelligent guidance services. The complex spatial correlation and nonlinear temporal dependence pose great challenges to obtaining accurate road conditions. Existing graph-based methods use a static adjacency matrix or a dynamic adjacency matrix to aggregate spatial information between nodes, which cannot fully represent the topological information. In this paper, we propose a Hybrid Graph Model (HGM) for accurate traffic prediction. The HGM constructs a static graph and a dynamic graph to represent the topological information of the traffic network, which is beneficial for mining potential and obvious spatial correlations. The proposed method combines a graph neural network, convolutional neural network, and attention mechanism to jointly extract complex spatial–temporal features. The HGM consists of two different sub-modules, called spatial–temporal attention module and dynamic graph convolutional network, to fuse complex spatial–temporal information. Furthermore, the proposed method designs a novel gated function to adaptively fuse the results from spatial–temporal attention and dynamic graph convolutional network to improve prediction performance. Extensive experiments on two real datasets show that the HGM outperforms comparable state-of-the-art methods. [ABSTRACT FROM AUTHOR]

To meet the increased customer demands, microgrid evolved. The structure of microgrid changes dynamically due to the intermittent nature of renewable-based generation, status of the distributed generator and opening of breakers for fault/maintenance. Hence, the magnitude of fault current is dynamic in nature. In order to deal with these dynamic changes, this paper addresses an adaptive central microgrid controller-based protection and relay coordination scheme, which revises the relay settings dynamically (both radial and looped configuration) for every change in topology. In the proposed algorithm, the primary relay responds to a fault immediately since the individual relays are given with fault level setting. For any abnormality in the network, the fault location is determined both via local relay and microgrid central controller (MCC). Hence, even though the local relay fails to identify the fault due to high fault impedance, the MCC locates the fault accurately and isolates the minimum faulty part. The coordination between relays is carried out by MCC in a time-graded manner based on microgrid central protection and relay coordination algorithm. The proposed algorithm is tested using Matlab in a microgrid built based on the IEEE 33 bus distribution network. [ABSTRACT FROM AUTHOR]

Let G = (V , E) be a simple connected graph with vertex set V and edge set E, respectively. The term "anti-reciprocal eigenvalue property" refers to a non-singular graph G for which, − 1 λ ∈ σ (G) , whenever λ ∈ σ (G) , ∀ λ ∈ σ (G) . Here, σ (G) is the multiset of all eigenvalues of A (G) . Moreover, if multiplicities of eigenvalues and their negative reciprocals are equal, then that graph is said to have strong anti-reciprocal eigenvalue properties, and the graph is referred to as a strong anti-reciprocal graph (or (− S R) graph). In this article, a new family of graphs F n (k , j) is introduced and the energy of F 5 (k , k 2) k ≥ 2 is calculated. Furthermore, with the help of F 5 (k , k 2) , some families of (− S R) graphs are constructed. [ABSTRACT FROM AUTHOR]

Applying machine learning algorithms to graph-structured data has garnered significant attention in recent years due to the prevalence of inherent graph structures in real-life datasets. However, the direct application of traditional deep learning algorithms, such as Convolutional Neural Networks (CNNs), is limited as they are designed for regular Euclidean data like 2D grids and 1D sequences. In contrast, graph-structured data are in a non-Euclidean form. Graph Neural Networks (GNNs) are specifically designed to handle non-Euclidean data and make predictions based on connectivity rather than spatial structure. Real-life graph data can be broadly categorized into two types: spatially-invariant graphs, where the link structure between nodes is independent of their spatial positions, and spatially-variant graphs, where node positions provide additional information about the graph's properties. However, there is limited understanding of the effect of spatial variance on the performance of Graph Neural Networks. In this study, we aim to address this issue by comparing the performance of GNNs and CNNs on spatially-variant and spatially-invariant graph data. In the case of spatially-variant graphs, when represented as adjacency matrices, they can exhibit Euclidean-like spatial structure. Based on this distinction, we hypothesize that CNNs may outperform GNNs when working with spatially-variant graphs, while GNNs may excel on spatially-invariant graphs. To test this hypothesis, we compared the performance of CNNs and GNNs under two scenarios: (i) graphs in the training and test sets had the same connectivity pattern and spatial structure, and (ii) graphs in the training and test sets had the same connectivity pattern but different spatial structures. Our results confirmed that the presence of spatial structure in a graph allows for the effective use of CNNs, which may even outperform GNNs. Thus, our study contributes to the understanding of the effect of spatial graph structure on the performance of machine learning methods and allows for the selection of an appropriate algorithm based on the spatial properties of the real-life graph dataset. [ABSTRACT FROM AUTHOR]

This note is intended as a contribution to the study of quantitative measures of graph complexity that use entropy measures based on symmetry. Determining orbit sizes of graph automorphism groups is a key part of such studies. Here we focus on an extreme case where every orbit contains just a single vertex. A permutation of the vertices of a graph G is an automorphism if, and only if, the corresponding permutation matrix commutes with the adjacency matrix of G. This fact establishes a direct connection between the adjacency matrix and the automorphism group. In particular, it is known that if the eigenvalues of the adjacency matrix of G are all distinct, every non-trivial automorphism has order 2. In this note, we add a condition to the case of distinct eigenvalues that makes the graph asymmetric, i.e., reduces the automorphism group to the identity alone. In addition, we prove analogous results for the Google and Laplacian matrices. The condition is used to build an O(n3) algorithm for detecting identity graphs, and examples are given to demonstrate that it is sufficient, but not necessary. [ABSTRACT FROM AUTHOR]

Community structure is a significant characteristic of complex networks, and community detection has valuable applications in network structure analysis. Non-negative matrix factorization (NMF) is a key set of algorithms used to solve the community detection issue. Nevertheless, the localization of feature vectors in the adjacency matrix, which represents the characteristics of complex network structures, frequently leads to the failure of NMF-based approaches when the data matrix has a low density. This paper presents a novel algorithm for detecting sparse network communities using non-negative matrix factorization (NMF). The algorithm utilizes local feature vectors to represent the original network topological features and learns regularization matrices. The resulting feature matrices effectively reveal the global structure of the data matrix, demonstrating enhanced feature expression capabilities. The regularized data matrix resolves the issue of localized feature vectors caused by sparsity or noise, in contrast to the adjacency matrix. The approach has superior accuracy in detecting community structures compared to standard NMF-based community detection algorithms, as evidenced by experimental findings on both simulated and real-world networks. [ABSTRACT FROM AUTHOR]

The Rubik's Cube mechanism (RCM) is a kind of reconfigurable mechanism with multiple characteristics such as multiple configurations, variable topology, strong coupling, and reconfigurability. Crossover research on the RCM with mathematics, chemistry, cryptography, and other disciplines has led to important breakthroughs and progress. It is obvious that the invention and creation of a new RCM can provide important ideological inspiration and theoretical guidance for the accelerated iterative updating of Rubik's Cube products and the expansion of their applications. This paper investigates the type synthesis method for a regular hexagonal prism (RHP) RCM (RHPRCM). Through analysis of the reconfigurable movement process of the RCM, two mechanism factors are abstracted, a type synthesis process for the RHPRCM is proposed, a symmetry layout method for the RCM's revolute axis based on the RHP space polyhedron is proposed, and an analysis method for the intersection of the revolute pair contact surfaces (RPCSs) based on the adjacency matrix is proposed. Taking a revolute axis passing through the center of an RHP and having only one RPCS for each revolute axis as an example, an RHPRCM with different topological structures is synthesized. The relevant research in this paper can provide methodological guidance for the synthesis of other spatial RCMs. [ABSTRACT FROM AUTHOR]

Video summarization aims to identify important segments in a video and merge them into a concise representation, enabling users to comprehend the essential information without watching the entire video. Graph structure-based video summarization approaches ignore the issue of redundant adjacency matrix. To address this issue, this paper proposes a video summary generation model based on graph structure reconstruction (VOGNet), in which the model first adopts a variational graph auto-encoders (VGAE) to reconstruct the graph structure to remove redundant information in the graph structure; followed by using the reconstructed graph structure in a graph attention network (GAT), allocating different weights to different shot features in the neighborhood; and lastly, in order to avoid the loss of information during the training of the model, a feature fusion approach is proposed to combine the training obtained shot features with the original shot features as the shot features for generating the summary. We perform extensive experiments on two standard datasets, SumMe and TVSum, and the experimental results demonstrate the effectiveness and robustness of the proposed model. [ABSTRACT FROM AUTHOR]

Unstructured Chinese medical texts are rich sources of entity and relational information. The extraction of entity relationships from medical texts is pivotal for the construction of medical knowledge graphs and aiding healthcare professionals in making swift and informed decisions. However, the extraction of entity relationships from these texts presents a formidable challenge, notably due to the issue of overlapping entity relationships. This study introduces a novel extraction model that leverages RoFormer's rotational position encoding (RoPE) technique for an efficient implementation of relative position encoding. This approach not only optimizes positional information utilization but also captures syntactic dependency information by constructing a weighted adjacency matrix. During the feature fusion phase, the model employs an entity attention mechanism for a deeper integration of features, effectively addressing the challenge of overlapping entity relationships. Experimental outcomes demonstrate that our model achieves an F1 score of 83.42 on datasets featuring overlapping entity relations, significantly outperforming other baseline models. [ABSTRACT FROM AUTHOR]

The timely detection of falls and alerting medical aid is critical for health monitoring in elderly individuals living alone. This paper mainly focuses on issues such as poor adaptability, privacy infringement, and low recognition accuracy associated with traditional visual sensor-based fall detection. We propose an infrared video-based fall detection method utilizing spatial-temporal graph convolutional networks (ST-GCNs) to address these challenges. Our method used fine-tuned AlphaPose to extract 2D human skeleton sequences from infrared videos. Subsequently, the skeleton data was represented in Cartesian and polar coordinates and processed through a two-stream ST-GCN to recognize fall behaviors promptly. To enhance the network's recognition capability for fall actions, we improved the adjacency matrix of graph convolutional units and introduced multi-scale temporal graph convolution units. To facilitate practical deployment, we optimized time window and network depth of the ST-GCN, striking a balance between model accuracy and speed. The experimental results on a proprietary infrared human action recognition dataset demonstrated that our proposed algorithm accurately identifies fall behaviors with the highest accuracy of 96%. Moreover, our algorithm performed robustly, identifying falls in both near-infrared and thermal-infrared videos. [ABSTRACT FROM AUTHOR]

To identify the complex time patterns in the process data and monitor the effect of wastewater treatment by predicting effluent chemical oxygen demand more accurately, a soft-sensor modeling method based on the multi-scale temporal convolutional network (MSTCN) was proposed in this paper. Data at different time scales are reconstructed according to the main frequencies determined by the Fourier transform approach, and the correlations between variables during that period are calculated and stored in the corresponding adjacency matrix. The specific temporal convolutional network (TCN) is built to learn the temporal dependencies within each sequence at the current scale, while the graph convolutional layer (GCN) captures the relationships among variables. Finally, predictions with less error can be obtained by integrating output features from GCN and TCN layers. The proposed model is validated on an annual dataset collected from a wastewater treatment plant employing biological processes for organic matter removal. The experimental results indicate that the proposed MSTCN reduces RMSE by 35.71% and 22.56% compared with the convolutional neural network and TCN, respectively. Moreover, MSCTN shortens the training period by 6.3 s and improves RMSE by 30.41% when compared to the long short-term memory network, which is outperformed in extracting temporal dynamic characteristics. [ABSTRACT FROM AUTHOR]