Dependency Relations Research Articles

Salient event detection via hypergraph convolutional network with cross-view self-supervised learning

Identifying the most dominant and central events in a document is critical for holistically understanding its important information. To measure the importance of an event, it is critical to understand its context: who is involved, where it happened, which other events it is related to, and what kind of relationship it has, among other factors. Although existing studies have achieved some accomplishments, they are still not fully effective for two main reasons: (1) They incorporate only the discrete global features ofthe document, which is insufficient for effectively capturing the contextual information of events; (2) They inadequately model the dependency relationships between events. According to previous research findings, it has been shown that hypergraphs effectively capture the global context (i.e., the document-level context) of long text. However, their potential for salient event detection has remained unexplored. To address this, we propose a novel framework called Salient Event Detection via Hypergraph Convolutional Network with Graph Self-supervised Learning (SEDGS). More specifically, we first construct two hypergraphs: one in the event argument view and another in the event view. We then propose a hypergraph convolutional network to model the event context and discourse relations between events. Moreover, to enhance hypergraph modeling and ensure consistency between argument-view and event-view event representations, we employ contrastive self-supervised learning (SSL) in our model training. Experimental results on a standard event salience dataset verify the superiority of SEDGS, advancing state-of-the-art models.

Multi-scale convolution enhanced transformer for multivariate long-term time series forecasting

In data analysis and forecasting, particularly for multivariate long-term time series, challenges persist. The Transformer model in deep learning methods has shown significant potential in time series forecasting. The Transformer model’s dot-product attention mechanism, however, due to its quadratic computational complexity, impairs training and forecasting efficiency. In addition, the Transformer architecture has limitations in modeling local features and dealing with multivariate cross-dimensional dependency relationship. In this article, a Multi-Scale Convolution Enhanced Transformer model (MSCformer) is proposed for multivariate long-term time series forecasting. As an alternative to modeling the time series in its entirety, a segmentation strategy is designed to convert the input original series into segmented forms with different lengths, then process time series segments using a new constructed multi-Dependency Aggregation module. This multi-Scale segmentation approach reduces the computational complexity of the attention mechanism part in subsequent models, and for each segment of length corresponds to a specific time scale, it also ensures that each segment retains the semantic information of the data sequence level, thereby comprehensively utilizing the multi-scale information of the data while more accurately capturing the real dependency of the time series. The Multi-Dependence Aggregate module captures both cross-temporal and cross-dimensional dependencies of multivariate long-term time series and compensates for local dependencies within the segments thereby captures local series features comprehensively and addressing the issue of insufficient information utilization. MSCformer synthesizes dependency information extracted from various temporal segments at different scales and reconstructs future series using linear layers. MSCformer exhibits higher forecasting accuracy, outperforming existing methods in multiple domains including energy, transportation, weather, electricity, disease and finance.

Recognition of Chinese Electronic Medical Records for Rehabilitation Robots: Information Fusion Classification Strategy.

Named entity recognition is a critical task in the electronic medical record management system for rehabilitation robots. Handwritten documents often contain spelling errors and illegible handwriting, and healthcare professionals frequently use different terminologies. These issues adversely affect the robot's judgment and precise operations. Additionally, the same entity can have different meanings in various contexts, leading to category inconsistencies, which further increase the system's complexity. To address these challenges, a novel medical entity recognition algorithm for Chinese electronic medical records is developed to enhance the processing and understanding capabilities of rehabilitation robots for patient data. This algorithm is based on a fusion classification strategy. Specifically, a preprocessing strategy is proposed according to clinical medical knowledge, which includes redefining entities, removing outliers, and eliminating invalid characters. Subsequently, a medical entity recognition model is developed to identify Chinese electronic medical records, thereby enhancing the data analysis capabilities of rehabilitation robots. To extract semantic information, the ALBERT network is utilized, and BILSTM and MHA networks are combined to capture the dependency relationships between words, overcoming the problem of different meanings for the same entity in different contexts. The CRF network is employed to determine the boundaries of different entities. The research results indicate that the proposed model significantly enhances the recognition accuracy of electronic medical texts by rehabilitation robots, particularly in accurately identifying entities and handling terminology diversity and contextual differences. This model effectively addresses the key challenges faced by rehabilitation robots in processing Chinese electronic medical texts, and holds important theoretical and practical value.

Automatic Aircraft Identification with High Precision from SAR Images Considering Multiscale Problems and Channel Information Enhancement

The SAR system possesses the ability to carry out all-day and all-weather imaging, which is highly valuable in the application of aircraft identification. However, aircraft identification from SAR images still faces great challenges due to speckle noise interference, multiscale problems, and complex background interference. To solve these problems, an efficient bidirectional path multiscale fusion and attention network (EBMA-Net) is proposed in this paper. It employs bidirectional connectivity to fuse the features of aircraft with different scales to perform the accurate detection of aircraft even when the background is highly complex. In the presented EBMA-Net, a module called efficient multiscale channel attention fusion (EMCA) and three parallel squeeze efficient channel attention (SECA) modules are proposed. In the EMCA module, the bidirectional paths are created by stacking upper and lower fusion modules, which effectively integrate shallow detailed features and deep semantic information. So, the detection performance of aircraft at different scales is improved. In the SECA module, the dependency relationship between feature channels is explicitly modeled, which can automatically learn the importance of different channels, prioritize key features, so as to improve the precision and robustness of aircraft identification. In the experiment, the public dataset of aircraft identification (i.e., SAR-AIRcraft-1.0, which is generated from the GF-3 satellite) from high-resolution SAR systems is used, and several other excellent target-detection networks are used for performance comparison, namely, YOLOv5s, YOLOv7, MGCAN, and EBPA2N. According to the results, the average aircraft detection accuracy of EBMA-Net is 91.31%, which is 4.5% higher than YOLOv7; and the false alarm rate is decreased by 5%. Its accuracy in the identification of aircraft can reach 95.6%, which is about 3.7% higher than YOLOv7. Therefore, the EBMA-Net obviously outperforms the other networks for aircraft detection and identification. The proposed EBMA-Net, which can capture the detailed information and better restrain the background interference, could also be used to perform the detection and identification of dense targets with different scales and background from SAR images.

Analysing the capacity of multispectral indices to map the spatial distribution of potential post-fire soil losses based on soil burn severity

The area burned in Spain exceeded historical records in 2022, when exceptionally warm conditions influenced wildfire events. The predicted intensification of wildfire regimes includes an increase in frequency, severity, and size. Therefore, a study of the wildfires that occurred in 2022 is necessary to understand their behaviour and possible environmental impacts. The objective of this study is to analyse the applicability of using spectral indices and Geographic Information System (GIS) approaches to map the spatial distribution and estimate potential soil losses using Sentinel-2 imagery and fire severity field data. Soil losses were estimated using an empirical model based on soil burn severity data collected in the field after wildfire. The relationship between the Normalized Difference Infrared Index (NDII), Difference Normalized Wildfire Ash Index (dNWAI), and the Blue Normalized Difference Vegetation Index (BNDVI) with the estimated soil losses was then evaluated. In addition, the influence of different time scales of the satellite images was analysed. The first period considered (Date I) ranges from 8 to 20 days after the beginning of the wildfire, which coincides with the field data collection. The second period considered (Date II) ranges from 28 to 35 days after the start of the wildfire. The results obtained showed a significant dependence relationship between the BNDVI index (using satellite images of Date I) and the estimated soil losses (R2 = 0.756), while the results of the NDII (R2 = 0.31) and dNWAI (R2 = 0.061), showed no spatial relationship with the estimated soil losses. Three of the largest wildfires in 2022 in Spain were analysed, and the results showed strong correlations of BNDVI index for Folgoso do Courel (R2 = 0.808), for Carballeda de Valedorras (R2 = 0.906), and for Sierra de la Culebra (R2 = 0.939). In addition, these results allowed the mapping and quantification of potential soil losses in areas where fire severity was high, totalling ∼2,50,000 Mg ha−1 in Folgoso do Courel, ∼3,70,000 Mg ha−1 in Carballeda de Valdeorras, and ∼4,70,000 Mg ha−1 in Sierra de la Culebra. Moreover, BNDVI values for estimating soil loss vary by vegetation type, and there is a positive correlation between severity classes and the BNDVI index. This approach can inform post-fire land management decisions in future wildfires and could be applied to other regions.

OpenECAD: An efficient visual language model for editable 3D-CAD design

Computer-aided design (CAD) tools are utilized in the manufacturing industry for modeling everything from cups to spacecraft. These programs are complex to use and typically require years of training and experience to master. Structured and well-constrained 2D sketches and 3D constructions are crucial components of CAD modeling. A well-executed CAD model can be seamlessly integrated into the manufacturing process, thereby enhancing production efficiency. Deep generative models of 3D shapes and 3D object reconstruction models have garnered significant research interest. However, most of these models produce discrete forms of 3D objects that are not editable. Moreover, the few models based on CAD operations often have substantial input restrictions. In this work, we fine-tuned pre-trained models to create OpenECAD models (0.55B, 0.89B, 2.4B and 3.1B), leveraging the visual, logical, coding, and general capabilities of visual language models. OpenECAD models can process images of 3D designs as input and generate highly structured 2D sketches and 3D construction commands, ensuring that the designs are editable. These outputs can be directly used with existing CAD tools’ APIs to generate project files. To train our network, we created a series of OpenECAD datasets. These datasets are derived from existing public CAD datasets, adjusted and augmented to meet the specific requirements of vision language model (VLM) training. Additionally, we have introduced an approach that utilizes dependency relationships to define and generate sketches, further enriching the content and functionality of the datasets.

Social Network Public Opinion Analysis Using BERT-BMA in Big Data Environment

The existing social network public opinion analysis methods have problems such as poor semantic expression quality and weak detection ability in short texts. Therefore, a social network public opinion analysis method based on BERT-BMA is proposed. To normalize the comment text, the rumor text is initially transferred to a word vector matrix using the BERT (Bidirectional Encoder Representations from Transformer) model. The BiLSTM-based network architecture is subsequently employed to acquire the trace features of data transmission. Ultimately, this study employs the multi-head attention mechanism to extract feature information that is more significant in the analysis of online public opinion by mining the dependency relationships between users, resulting in increasing ability to detect public opinion emergencies. The experimental outcomes indicate that the results on the Twitter data set and Weibo dataset are superior to other comparative models.

A rigorous and versatile statistical test for correlations between stationary time series.

In disciplines from biology to climate science, a routine task is to compute a correlation between a pair of time series and determine whether the correlation is statistically significant (i.e., unlikely under the null hypothesis that the time series are independent). This problem is challenging because time series typically exhibit autocorrelation and thus cannot be properly analyzed with the standard iid-oriented statistical tests. Although there are well-known parametric tests for time series, these are designed for linear correlation statistics and thus not suitable for the increasingly popular nonlinear correlation statistics. There are also nonparametric tests that can be used with any correlation statistic, but for these, the conditions that guarantee correct false positive rates are either restrictive or unclear. Here, we describe the truncated time-shift (TTS) test, a nonparametric procedure to test for dependence between 2 time series. We prove that this test correctly controls the false positive rate as long as one of the time series is stationary, a minimally restrictive requirement among current tests. The TTS test is versatile because it can be used with any correlation statistic. Using synthetic data, we demonstrate that this test performs correctly even while other tests suffer high false positive rates. In simulation examples, simple guidelines for parameter choices allow high statistical power to be achieved with sufficient data. We apply the test to datasets from climatology, animal behavior, and microbiome science, verifying previously discovered dependence relationships and detecting additional relationships.

Research on transition sequence partitioning based on dependency relationships

Research on transition sequence partitioning based on dependency relationships

GNNctd: A graph neural network based on complicated temporal dependencies modeling for fashion trend prediction

Fashion trend forecasting has consistently remained a focal point within the realm of fashion. Existing methods predominantly concentrate on the external factors influencing fashion trends, often disregarding the intricate interplay among distinct fashion elements, namely the ‘spatial dependencies’ among them. It is also a significant challenge to excavate complicated temporal relationships in complex time series data. In this research, our primary focus is modeling the relationships among diverse fashion elements and the intricate temporal dependencies within time series data. First, we propose a Street Fashion Trend (SFT) dataset by leveraging images from the popular photo-sharing platform Flickr 11Flickr: www.flickr.com. and the renowned fashion show website Vogue.22Vogue: https://m.vogue.com.cn/. Furthermore, we propose a model called GNNctd to solve the above problems. This model leverages a spatial dependency capture module (SDCM) based on a graph neural network to dynamically model the ‘spatial dependency relationships’ among distinct fashion elements. Meanwhile, the model introduces a temporal relationship extraction block (TREB), which comprises two pivotal modules: the interaction learning (IL) module designed to capture local temporal dependencies and a global time attention module (GTAM) that is used to capture global temporal dependencies. Many experiments substantiate that the proposed GNNctd model can achieve more accurate predictions of fashion trends in our constructed dataset and the other three fashion trend datasets. Simultaneously, the GNNctd model achieves a state-of-the-art performance on the Solar-Energy, Exchange Rate, and Wind datasets within the domain of time series prediction.

Remaining useful life prediction for a cracked rotor system via moving feature fusion based deep learning approach

This study proposes a novel approach for predicting the remaining useful life (RUL) of a cracked rotor system by using a moving fusion gated recurrent unit (MFGRU) model. Firstly, 12 multi-domain features were extracted from the raw collected vibration signals, which were then fused and distributed adaptively using bidirectional exponential moving average (EMA) and multi-head attention (MHA). Then a bidirectional gated recurrent unit network combined with moving feature fusion method was proposed to capture the long-term dependence relationship in the monitoring data of the cracked rotor for RUL prediction. Finally, the model’s performance was validated through accelerated life experiments, with the measured values of root mean square error (RMSE) and mean absolute error (MAE) below 3.17 and 2.60, respectively. This study offers a dynamic RUL prediction method with certain significance and valuable reference for designing deep learning models.

A novel fatigue driving detection method based on whale optimization and Attention-enhanced GRU

Fatigue driving has emerged as the predominant causative factor for road traffic safety accidents. The fatigue driving detection method, derived from laboratory simulation data, faces challenges related to imbalanced data distribution and limited recognition accuracy in practical scenarios. In this study, we introduce a novel approach utilizing a gated recurrent neural network method, employing whale optimization algorithm for fatigue driving identification. Additionally, we incorporate an attention mechanism to enhance identification accuracy. Initially, this study focuses on the driver's operational behavior under authentic vehicular conditions. Subsequently, it employs wavelet energy entropy, scale entropy, and singular entropy analysis to extract the fatigue-related features from the driver's operational behavior. Subsequently, this study adopts the cross-validation recursive feature elimination method to derive the optimal fatigue feature index about operational behavior. To effectively capture long-range dependence relationships, this study employs the gated recurrent unit neural network method. Lastly, an attention mechanism is incorporated in this study to concentrate on pivotal features within the data sequence of driving behavior. It assigns greater weight to crucial information, mitigating information loss caused by the extended temporal sequence. Experimental results obtained from real vehicle data demonstrate that the proposed method achieves an accuracy of 89.84% in third-level fatigue driving detection, with an omission rate of 10.99%. These findings affirm the feasibility of the approach presented in this study.

The application of the analysis framework for compound extreme event dependencies in China

AbstractThe escalation of compound extreme events has resulted in noteworthy economic and property losses. Recognizing the intricate interconnections among these events has become imperative. To tackle this challenge, we have formulated a comprehensive framework for the systematic analysis of their dependencies. This framework consists of three steps. (1) Define extreme events using Mahalanobis distance thresholds. (2) Represent dependencies among multiple extreme events through a point process‐based method. (3) Verify dependencies with residual tail coefficients, determining the final dependency structure. Applying this framework to assess the extreme dependence of precipitation on wind speed and temperature in China, revealed four distinct dependency structures. In northern, Jianghuai, and southern China, precipitation heavily relies on wind speed, while temperatures maintain relative independence. In northeastern and northwestern China, precipitation exhibits relative independence, yet a notable dependence exists between temperatures and wind speed. In southwestern China, precipitation strongly depends on temperature, while wind speed remains relatively independent. The Qinghai–Tibet Plateau region displays a significant dependence relationship among precipitation, wind speed, and temperature, with weaker dependence between extreme wind speed and temperature. This framework is instrumental for analyzing dependencies among extreme values in compound events.

A novel reliability analysis approach for multi‐component systems with stochastic dependency and functional relationships

A novel reliability analysis approach for multi‐component systems with stochastic dependency and functional relationships

Appearance-posture fusion network for distracted driving behavior recognition

In recent years, detection techniques using computer vision and deep learning have shown promise in assessing driver distraction. This paper proposes a fusion network that combines a Spatial-Temporal Graph Convolutional Network (ST-GCN) and a hybrid convolutional network to integrate multimodal input data for recognizing distracted driver behavior. Specifically, to address the limitations of the ST-GCN method in modeling long-distance joint interaction features and inadequate temporal feature extraction, we design the Spatially Symmetric Configuration Partitioning Graph Convolutional Network (SSCP-GCN) to model relative motion information of symmetric relationships between limbs. Specifically, we utilize densely connected blocks for processing multi-scale temporal information between consecutive frames, thereby enhancing the reuse of bottom features. Furthermore, the expression of important temporal information is augmented by the introduction of the channel attention mechanism. To tackle the problem that the Mixed Convolution (MC) combining 3D convolution with 2D convolution cannot extract higher-order timing information and has limitations in modeling global dependency relationships, we compensate for its inability to capture higher-order temporal semantic information using the Time Shift Module (TSM) without consuming additional computational resources. Additionally, the 3D Multi-Head Self-Attention mechanism (3D MHSA) is employed to integrate global spatial–temporal information of high-level features, avoiding the issue of model complexity proliferation caused by the deep stacking design of Convolutional Neural Networks (CNN). Lastly, we introduce a multistream network framework that integrates driver posture and appearance features to harness complementary advantages, enabling us to combine multimodal input features to achieve better model performance. Experimental results indicate that the accuracy of the network designed in this paper reaches 95.6% and 94.3% on ASU dataset and NTU-RGB+D dataset, respectively. The small size of the model offers the possibility for practical application of the algorithm.

SWFormer: A scale-wise hybrid CNN-Transformer network for multi-classes weed segmentation

Weeds in rapeseed field are an important factor in crop yield reduction and economic loss. Thus, Precision Agriculture is an important task for sustainable agriculture and weed management. At present, deep learning techniques have shown great potential for image-based detection and classification in various crops and weeds. However, the inherent limitations of traditional convolutional neural networks pose significant challenges due to the locally similarity of weeds and crops in color, shape and texture. To address this issue, we introduce SWFormer, a scale-wise hybrid CNN-Transformer network. SWFormer leverages the distinct strengths of both convolutional and transformer architectures. Convolutional structures excel at extracting short-range dependency information among pixels, whereas transformer structures are adept at capturing global dependency relationships. Additionally, we propose two innovative modules. Firstly, the Scale-wise Cascade Convolution (SWCC) module is designed to capture multiscale features and expand the receptive field. Secondly, the Adaptive Semantic Aggregation (ASA) module facilitates adaptive and effective information fusion across two distinct feature maps. Our experiments were conducted on the publicly available cropandweed dataset and SB20 dataset. it yields improved performance over other mainstream segmentation models. Specifically, SWFormer with 52.33M/527.51GFLOPs achieves an mAP of 76.54% and an accuracy of 83.95% on the cropandweed dataset. For the SB20 dataset, it attains an mAP of 61.24% and an accuracy of 79.47%. Overall, the evaluation clearly demonstrates our proposed SWFormer is conducive to promoting further research in the area of Precision Agriculture.

Road extraction from remote sensing images based on a multi-scale asymmetric dual attention mechanism

ABSTRACT Aiming at the problems of road fracture and detail loss caused by not considering the geometric features in the road extraction method. We proposed an encoder-decoder architecture based on a multi-scale asymmetric dual attention mechanism. Firstly, A multi-scale convolution block in the shape of ‘Union Jack’ is designed. It includes symmetric convolution and asymmetric convolution along horizontal, vertical, left diagonal, and right diagonal spatial directions, and a multi-scale dilated convolution for extracting features of different scales. Remote dependence relationships are highly converged by using it, and road fracture problems caused by occlusion can be solved effectively. Secondly, a directional dual attention mechanism is proposed, which consists of directional channel attention using strip pooling and a directional spatial attention mechanism using asymmetric convolution along left diagonal and right diagonal spatial directions. It can use the directivity of asymmetric convolution to allocate attention mechanism adaptively in attention mechanism, and effectively avoid the road detail loss problem. Finally, we conducted corresponding experiments on the DeepGlobe and Ottawa road datasets, and the experimental results are superior to the current state-of-the-art methods.

An Unavoidable Mind-Set Reversal: Consciousness in Vision Science.

In recent decades, the debate on consciousness has been conditioned by the idea of bottom-up emergence, which has influenced scientific research and raised a few obstacles to any attempt to bridge the explanatory gap. The analysis and explanation of vision conducted according to the accredited methodologies of scientific research in terms of physical stimuli, objectivity, methods, and explanation has encountered the resistance of subjective experience. Moreover, original Gestalt research into vision has generally been merged with cognitive neuroscience. Experimental phenomenology, building on the legacy of Gestalt psychology, has obtained new results in the fields of amodal contours and color stratifications, light perception, figurality, space, so-called perceptual illusions, and subjective space and time. Notwithstanding the outcomes and the impulse given to neuroscientific analyses, the research carried out around these phenomena has never directly confronted the issue of what it means to be conscious or, in other words, the nature of consciousness as self-referentiality. Research has tended to focus on the percept. Therefore, explaining the non-detachability of parts in subjective experience risks becoming a sort of impossible achievement, similar to that of Baron Munchausen, who succeeds in escaping unharmed from this quicksand by pulling himself out by his hair. This paper addresses how to analyze seeing as an undivided whole by discussing several basic dimensions of phenomenal consciousness on an experimental basis and suggesting an alternative way of escaping this quicksand. This mind-set reversal also sheds light on the organization and dependence relationships between phenomenology, psychophysics, and neuroscience.

The Tail Dependence and Lead-Lag Relationship in Financial Markets

The Tail Dependence and Lead-Lag Relationship in Financial Markets

A resource prediction method for air traffic cyber-physical-social system

Air traffic is exhibiting the characteristics of large flow, strong coupling, and high time variation. Therefore, the complex network of air traffic is more vulnerable to disturbances. When it is disturbed, the failure of some nodes spreads through dependency relationships in the network, resulting in cascade failure. In the event of a cascade failure, the network may quickly collapse until it is paralyzed, with widespread delays and flight cancellations. The current flow management and deployment methods still remain in the control-oriented stage, which is mainly completed by air traffic controls (ATCs), and lack of accurate flow adjustment and effective utilization of capacity. The whole air traffic system and its peripheral factors are intricate, so human and social factors must be integrated into the control and decision-making of the system. Considering engineering and social factors such as operation environment, social environment, personnel, rules, equipment, and information processing, we analyse the air traffic in a cyber-physical-social system (CPSS). To reflect the actual system behaviour rules, dynamic response, limit state, and so on, the corresponding computational experiment and comprehensive evaluation system are established. Based on neural networks and other technologies, a resource prediction scheme based on task demand is proposed for multi-dimensional resources such as airports, air routes, and ATC, to reduce the cost of system resource scheduling and improve resource utilization through resource prediction and adjustment. Finally, the accuracy of the proposed resource prediction algorithm is verified by theoretical analysis and simulation.