Search Results (13)

The cross-lingual text classification task remains a long-standing challenge that aims to train a classifier on high-resource source languages and apply it to classify texts in low-resource target languages, bridging linguistic gaps while maintaining accuracy. Most existing methods achieve exceptional performance by relying on multilingual pretrained language models to transfer knowledge across languages. However, little attention has been paid to factors beyond semantic similarity, which leads to the degradation of classification performance in the target languages. This study proposes a novel framework, a heterogeneous graph neural network with multi-view contrastive learning for cross-lingual text classification, which integrates a heterogeneous graph architecture with multi-view contrastive learning for the cross-lingual text classification task. This study constructs a heterogeneous graph to capture both syntactic and semantic knowledge by connecting document and word nodes using different types of edges, including Part-of-Speech tagging, dependency, similarity, and translation edges. A Graph Attention Network is applied to aggregate information from neighboring nodes. Furthermore, this study devises a multi-view contrastive learning strategy to enhance model performance by pulling positive examples closer together and pushing negative examples further apart. Extensive experiments show that the framework outperforms the previous state-of-the-art model, achieving improvements of 2.20% in accuracy and 1.96% in F1-score on the XGLUE and Amazon Review datasets, respectively. These findings demonstrate that the proposed model makes a positive impact on the cross-lingual text classification task overall. Full article

With the rapid evolution of intelligent driving technology, vehicle trajectory prediction has become a pivotal technique for enhancing road safety and traffic efficiency. In this domain, high-definition vector maps and graph neural networks (GNNs) play a vital role, supporting precise vehicle positioning and optimizing path planning, thereby improving the performance of intelligent driving systems. However, high-definition vector maps and traditional GNNs still encounter several challenges in trajectory prediction, such as high computational resource demands, long training times, and limited modeling capabilities for dynamic traffic environments and complex interactions. To address these challenges, this paper proposes an adaptive edge generator method, this method dynamically constructs and optimizes the connections between nodes in the GNN architecture, effectively enhancing the accuracy and efficiency of trajectory prediction. Specifically, we classify nodes into dynamic and static nodes based on their attributes, and devise differentiated edge construction strategies accordingly. For dynamic nodes, we introduce a relative angle factor, enabling the attention model to comprehensively consider the distance and intersection status between nodes, resulting in more accurate computation of edge weights. For static nodes, we utilize a length threshold to assess the feasibility of establishing connections between vehicles and lane lines, determining whether a connection should be established. Through this approach, we successfully reduce the algorithmic complexity, increase computational speed, and maintain high trajectory prediction accuracy. Tests on the Argoverse motion prediction dataset demonstrate that trajectory prediction utilizing the adaptive edge generator achieves an average displacement error (ADE) of 0.6681, a final displacement error (FDE) of 0.9864, and a miss rate (MR) of 0.0952. Furthermore, the model parameters are significantly reduced, validating the effectiveness of the proposed vehicle trajectory prediction method based on the adaptive edge generator. Full article

Link prediction is a key technique for connecting entities and relationships in a graph reasoning field. It leverages known information about the graph structure data to predict missing factual information. Previous studies have either focused on the semantic representation of a single triplet or on the graph structure data built on triples. The former ignores the association between different triples, and the latter ignores the true meaning of the node itself. Furthermore, common graph-structured datasets inherently face challenges, such as missing information and incompleteness. In light of this challenge, we present a novel model called Multi-source Information Graph Embedding with Ensemble Learning for Link Prediction (EMGE), which can effectively improve the reasoning of link prediction. Ensemble learning is systematically applied throughout the model training process. At the data level, this approach enhances entity embeddings by integrating structured graph information and unstructured textual data as multi-source information inputs. The fusion of these inputs is effectively addressed by introducing an attention mechanism. During the training phase, the principle of ensemble learning is employed to extract semantic features from multiple neural network models, facilitating the interaction of enriched information. To ensure effective model learning, a novel loss function based on contrastive learning is devised, effectively minimizing the discrepancy between predicted values and the ground truth. Moreover, to enhance the semantic representation of graph nodes in link prediction, two rules are introduced during the aggregation of graph structure information. These rules incorporate the concept of spreading activation, enabling a more comprehensive understanding of the relationships between nodes and edges in the graph. During the testing phase, the EMGE model is validated on three datasets, including WN18RR, FB15k-237, and a private Chinese financial dataset. The experimental results demonstrate a reduction in the mean rank (MR) by 0.2 times, an improvement in the mean reciprocal rank (MRR) by 5.9%, and an increase in the Hit@1 by 12.9% compared to the baseline model. Full article

The application of machine learning (ML) for the automatic classification of building elements is a powerful technique for ensuring information integrity in building information models (BIMs). Previous work has demonstrated the favorable performance of such models on classification tasks using geometric information. This research explores the hypothesis that incorporating contextual information into the ML models can improve classification accuracy. To test this, we created a graph data structure where each building element is represented as a node assigned with basic geometric information. The connections between the graph nodes (edges) represent the immediate neighbors of that node, capturing the contextual information expressed in the BIM model. We devised a process for extracting graphs from BIM files and used it to construct a graph dataset of over 42,000 building elements and used the data to train several types of ML models. We compared the classification results of models that rely only on geometry, to graph neural networks (GNNs) that leverage contextual information. This work demonstrates that graph-based models for building element classification generally outperform classic ML models. Furthermore, dividing the graphs that represent complete buildings into smaller subgraphs further improves classification accuracy. These results underscore the potential of leveraging contextual information via graphs for advancing ML capabilities in the BIM environment. Full article

Identifying the changes in landscape pattern and ecosystem service value (ESV) and clarifying their relationship in temporal changes and spatial variations can provide insight into regional landscape features and scientific support for regional landscape planning. Leveraging land use data from the Yihe River Basin, we quantitatively assessed the landscape pattern and ESV shifts spanning from 2000 to 2018 using the landscape pattern indexes and the equivalence factor method. We employed Pearson correlation metrics and the geographically weighted regression model to explore the interrelation of their spatiotemporal variations. Our results show the following: (1) Forestland represents the most expansive land cover category. Apart from construction land, all other types experienced a decline in area. The most notable change occurred in the area of construction land. (2) The aggregation of the overall landscape shows a downward trend. The levels of fragmentation, landscape diversity, and richness increased. (3) Throughout the entire study period, the overall ESV gradually decreased, and the land cover type with the greatest contribution to the ESV was forestland. (4) In terms of temporal changes, the patch density and edge density of the overall area are significantly negatively correlated with total ESVs. The largest values for the patch index, perimeter–area fractal dimension (PAFRAC), and aggregation are significantly positively correlated with total ESVs. (5) In terms of spatial variation, the contagion index (CONTAG), PAFRAC, and the Shannon diversity index (SHDI) were noticeably correlated with ESVs. The CONTAG is positively correlated with ESVs upstream, but negatively midstream and downstream. The SHDI is negatively correlated with ESVs upstream, but positively midstream and downstream. The PAFRAC exhibits a positive correlation with ESVs for the most part. The association between the landscape pattern indexes and ESVs exhibits temporal and spatial inconsistencies in most instances, suggesting a spatiotemporal scale effect in their relationship. This study recommends that the local government devises a long-term strategy for urban development and exercises stringent control over the unregulated expansion of construction land. Through reasonable territorial spatial planning, government departments could enhance the connectivity of the overall landscape pattern of the Yihe River Basin to achieve the reasonable allocation and sustainable development of regional resources. Full article

In recent years, technological advancements in sensor, communication, and data storage technologies have led to the increasingly widespread use of smart devices in different types of buildings, such as residential homes, offices, and industrial installations. The main benefit of using these devices is the possibility of enhancing different crucial aspects of life within these buildings, including energy efficiency, safety, health, and occupant comfort. In particular, the fast progress in the field of the Internet of Things has yielded exponential growth in the number of connected smart devices and, consequently, increased the volume of data generated and exchanged. However, traditional Cloud-Computing platforms have exhibited limitations in their capacity to handle and process the continuous data exchange, leading to the rise of new computing paradigms, such as Edge Computing and Fog Computing. In this new complex scenario, advanced Artificial Intelligence and Machine Learning can play a key role in analyzing the generated data and predicting unexpected or anomalous events, allowing for quickly setting up effective responses against these unexpected events. To the best of our knowledge, current literature lacks Deep-Learning-based approaches specifically devised for guaranteeing safety in IoT-Based Smart Buildings. For this reason, we adopt an unsupervised neural architecture for detecting anomalies, such as faults, fires, theft attempts, and more, in such contexts. In more detail, in our proposal, data from a sensor network are processed by a Sparse U-Net neural model. The proposed approach is lightweight, making it suitable for deployment on the edge nodes of the network, and it does not require a pre-labeled training dataset. Experimental results conducted on a real-world case study demonstrate the effectiveness of the developed solution. Full article

The application of edge computing combined with the Internet of Things (edge-IoT) has been rapidly developed. It is of great significance to develop a lightweight network for gearbox compound fault diagnosis in the edge-IoT context. The goal of this paper is to devise a novel and high-accuracy lightweight neural network based on Legendre multiwavelet transform and multi-channel convolutional neural network (LMWT-MCNN) to fast recognize various compound fault categories of gearbox. The contributions of this paper mainly lie in three aspects: The feature images are designed based on the LMWT frequency domain and they are easily implemented in the MCNN model to effectively avoid noise interference. The proposed lightweight model only consists of three convolutional layers and three pooling layers to further extract the most valuable fault features without any artificial feature extraction. In a fully connected layer, the specific fault type of rotating machinery is identified by the multi-label method. This paper provides a promising technique for rotating machinery fault diagnosis in real applications based on edge-IoT, which can largely reduce labor costs. Finally, the PHM 2009 gearbox and Paderborn University bearing compound fault datasets are used to verify the effectiveness and robustness of the proposed method. The experimental results demonstrate that the proposed lightweight network is able to reliably identify the compound fault categories with the highest accuracy under the strong noise environment compared with the existing methods. Full article

Using the Precolumbian lowland Maya model of urban soil connectivity discussed in Part I, we review how soil connectivity can transition into urban planning policy and, by extension, could ultimately become codified as vantages and guidelines for urban design. In Maya agro-urban landscapes, the interspersion of open and green space with construction and paving provides edges (or interfaces) between sealed and unsealed soils at which the potential for soil connectivity manifests. These edges create an undeniable opportunity for urban planning to determine methods, guidelines, and conditions that can enhance soil connectivity. We argue that adequate attention to soils in urban sustainability goals would counteract misconceptions about the compact city paradigm and compensation for soil sealing in urban practice. Through preserving and increasing urban soil availability, proximity, and accessibility, advisory policies can stimulate shared values and everyday behaviours that reinforce the responsible and productive use of urban soils. Such urban planning can enable and encourage widespread participation in urban soil management. To promote policymaking on urban soils, we assess the importance and challenges of using urban green space as a proxy for the presence of urban soils. Our review suggests that urban green space offers high potential for use in urban planning to develop habit architectures that nurture soil-oriented pro-environmental behaviour. However, we also acknowledge the need for consistent and systematic data on urban soils that match sustainable urban development concepts to assist the effective transition of soil connectivity into urban planning codifications. Formulating adequate soil-oriented planning guidelines will require translating empirical insights into policy applications. To this end, we propose methods for enhancing our understanding and ability to monitor urban soil connectivity, including onsite surveys of land-use and bottom-up experience of soils, the mapping of the edges between sealed and unsealed soils, and using landscape ecological scales of analysis. In conclusion, we position soil care and connectivity as a primary task for urban planning and design and digest our findings and empirical vantages into concrete starting points devised as instruments to support urban planning in achieving soil codification. Full article

Elucidating the design principles of regulatory networks driving cellular decision-making has important implications for understanding cell differentiation and guiding the design of synthetic circuits. Mutually repressing feedback loops between ‘master regulators’ of cell fates can exhibit multistable dynamics enabling “single-positive” phenotypes: (high A, low B) and (low A, high B) for a toggle switch, and (high A, low B, low C), (low A, high B, low C) and (low A, low B, high C) for a toggle triad. However, the dynamics of these two motifs have been interrogated in isolation in silico, but in vitro and in vivo, they often operate while embedded in larger regulatory networks. Here, we embed these motifs in complex larger networks of varying sizes and connectivity to identify hallmarks under which these motifs maintain their canonical dynamical behavior. We show that an increased number of incoming edges onto a motif leads to a decay in their canonical stand-alone behaviors. We also show that this decay can be exacerbated by adding self-inhibition but not self-activation loops on the ‘master regulators’. These observations offer insights into the design principles of biological networks containing these motifs and can help devise optimal strategies for the integration of these motifs into larger synthetic networks. Full article

The expected huge amount of connected cars and applications with varying Quality of Service (QoS) demands still depend on agile/flexible networking infrastructure to deal with dynamic service requests to the control plane, which may become a bottleneck for 5G and Beyond Software-Defined Network (SDN) based Internet of Vehicles (IoV). At the heart of this issue is the need for an architecture and optimization mechanisms that benefit from cutting edge technologies while granting latency bounds in order to control and manage the dynamic nature of IoV. To this end, this article proposes an autonomic software-defined vehicular architecture grounded on the synergy of Multi-access Edge Computing (MEC) and Network Functions Virtualization (NFV) along with a heuristic approach and an exact model based on linear programming to efficiently optimize the dynamic resource allocation of SDN controllers, ensuring load balancing between controllers and employing reserve resources for tolerance in case of demand variation. The analyses carried out in this article consider: (a) to avoid waste of limited MEC resources, (b) to devise load balancing among controllers, (c) management complexity, and (d) to support scalability in dense IoV scenarios. The results show that the heuristic efficiently manages the environment even in highly dynamic and dense scenarios. Full article

In this paper, we present initial results from our distributed edge systems research in the domain of sustainable harvesting of common good resources in the Arctic Ocean. Specifically, we are developing a digital platform for real-time privacy-preserving sustainability management in the domain of commercial fishery surveillance operations. This is in response to potentially privacy-infringing mandates from some governments to combat overfishing and other sustainability challenges. Our approach is to deploy sensory devices and distributed artificial intelligence algorithms on mobile, offshore fishing vessels and at mainland central control centers. To facilitate this, we need a novel data plane supporting efficient, available, secure, tamper-proof, and compliant data management in this weakly connected offshore environment. We have built our first prototype of Dorvu, a novel distributed file system in this context. Our devised architecture, the design trade-offs among conflicting properties, and our initial experiences are further detailed in this paper. Full article

The world is evolving, and it has transformed from the industrial age to the era of connected and intelligent products in both organizations and competition. The advances in technology in the last decade have led to the introduction of a new term called Industry 4.0 or the fourth industrial revolution and that has led to the emergence of the term Quality 4.0. Quality 4.0 is the digitalization of traditional quality approaches and the focus on the use of digital tools to improve an organization’s ability to meet customers’ requirements with high quality. The purpose of this paper is to assess the environments of higher education institutions (HEIs) against the 11 axes of LNS Research Quality 4.0 framework and provide insights about their readiness for Quality 4.0 transformation. The framework helps the organizations digitalize their traditional quality practices and transform to Quality 4.0 through exploring the traditional quality—Quality 4.0 continuum of tools and/or concepts related to each axis so they can assess their transformation efforts accordingly. This paper uses these continuums to identify the quality implementation efforts conducted by HEIs through analyzing the continuums’ related practices adopted within their environments and find out what should be done to get to the full transformation to Quality 4.0 within the higher education field. The study shows the HEIs potential of adopting the Quality 4.0 tools and techniques of varies axes of the framework while revealing a limited adoption of most of them in the current times. This is due to several challenges the most impacting of which is having fragmented processes together with fragmented data systems and sources. The study is concluded with a proposed roadmap to assist HEIs to get the best out their efforts in the Quality 4.0 transformation process. Full article

Shadows exist universally in sunlight-source remotely sensed images, and can interfere with the spectral morphological features of green vegetations, resulting in imprecise mathematical algorithms for vegetation monitoring and physiological diagnoses; therefore, research on shadows resulting from forest canopy internal composition is very important. Red edge is an ideal indicator for green vegetation’s photosynthesis and biomass because of its strong connection with physicochemical parameters. In this study, red edge parameters (curve slope and reflectance) and the normalized difference vegetation index (NDVI) of two species of coniferous trees in Inner Mongolia, China, were studied using an unmanned aerial vehicle’s hyperspectral visible-to-near-infrared images. Positive correlations between vegetation red edge slope and reflectance with different illuminated/shaded canopy proportions were obtained, with all R²s beyond 0.850 (p < 0.01). NDVI values performed steadily under changes of canopy shadow proportions. Therefore, we devised a new vegetation index named normalized difference canopy shadow index (NDCSI) using red edge’s reflectance and the NDVI. Positive correlations (R² = 0.886, p < 0.01) between measured brightness values and NDCSI of validation samples indicated that NDCSI could differentiate illumination/shadow circumstances of a vegetation canopy quantitatively. Combined with the bare soil index (BSI), NDCSI was applied for linear spectral mixture analysis (LSMA) using Sentinel-2 multispectral imaging. Positive correlations (R² = 0.827, p < 0.01) between measured brightness values and fractional illuminated vegetation cover (FIVC) demonstrate the capacity of NDCSI to accurately calculate the fractional cover of illuminated/shaded vegetation, which can be utilized to calculate and extract the illuminated vegetation canopy from satellite images. Full article