Search

Presence can promote online learners' learning effectiveness in higher education, but in livestream teaching, the influential relationship between different types of presence and learning effectiveness is unclear. Therefore, based on the Community of Inquiry (CoI) Framework, we used structural equation, hierarchical regression, and the Bootstrap self-serving method to conduct a survey on college students participating in livestream teaching practice. The research findings revealed that livestream teaching substantially impacts learning effectiveness, with teaching presence, social presence, and cognitive presence all playing crucial roles. Notably, teaching presence has a significant positive influence on learning effectiveness through two key mediating factors: Social presence and cognitive presence. Consequently, three distinct mediating paths are identified. Among these three mediating paths, the most optimal route for teaching presence to enhance learning effectiveness is mediating cognitive presence. In conclusion, we recommend improving the livestream teaching environment, guiding learners toward active participation to promote a sense of embodiment, and elaborately designing livestream learning activities to improve interactivity. Finally, this paper offers evidence and insights for the improvement of livestream teaching in colleges, which will enhance learners' overall learning effectiveness.

IntroductionWith the acceleration of urbanization, human population and built surface in urban areas have increased rapidly, triggering numerous environmental problems. Identification of ecological security pattern (ESP) can be helpful to optimize the interaction and relationship between ecological conservation and socioeconomic development in a given region. In this study, taking Chengdu City as an example, a methodology was used to analyze the city’s ESP.MethodologyEcological sources were identified based on their ecological security values, which were derived from the spatio-temporal evaluation of ecosystem services and ecological health. The ecological resistance surface was revised with nighttime light index. Linkage Mapper combined with circuit theory was used to extracted ecological corridors and key ecological nodes.Results:The results showed that the mean values of ecological security in Chengdu City presented a downward and then upward trend from 2000 to 2018. In 2018, the ESP in Chengdu City was formed by 140 ecological sources with the total area of 8,819.78 km2, 302 ecological corridors with the area of 456.91 km2, as well as 61 pinch-points and 17 barrier points. The ecological sources in Chengdu City were mainly distributed in or nearby Longmen Mountain and Longquan Mountain, two flanks of the Chengdu Plain, and connected each other with hundreds of ecological corridors. Most of ecological sources and corridors were composited of forests in mountainous areas and linpan units (wooded lots) in rural plain areas. On the contrary, pinch-points and barrier points were mainly distributed in the districts nearby the metropolitan urban center.ConclusionFinally, this study proposed that constructing ESP must adopt a dynamic and holistic approach in considering the change of restoring ecosystems, shift of urban demands for ecosystem services and transformation of urban-rural land use/cover. Restoring and managing corridors sustainably are important for improving the ecological connectivity of all over ecological sources in a city, all of which together, if functioning well, could further support the realization of sustainable urbanization.

Solving arithmetic word problems ($ AWPs $) that involve deep implicit relations can be quite challenging. However, the paper proposed two approaches to tackle this issue. The first approach used the modifier-to-matrix ($ MTM $) model to extract noun modification components from the problem text. Specifically, a missing entity recovery ($ MER $) model translated explicit expressions into a node dependency graph ($ NDG $). The nodes on the graph then recursively acquired connections from the knowledge base through the $ MER $ model until the goal was achieved with a known quantity. The solving engine then selected the appropriate knowledge as the prompt. The second approach proposed in the paper was a comprehensive one that combined explicit and implicit knowledge to enhance reasoning abilities. The experimental results of the dataset demonstrate that the proposed algorithm is superior to the baseline algorithms in solving $ AWPs $ that require deep implicit relations.

Arbuscular mycorrhizal fungi (AMF) play an indispensable role in terrestrial ecosystem soil organic carbon (SOC) dynamics. Over the past 22 years, numerous researchers have studied the role of AMF on soil the carbon cycle, thereby producing a wealth of knowledge. However, a comprehensive summary and analysis of the basic characteristics, research outputs, and knowledge base is still lacking. To explore the authors, countries, institutions, and keywords involved in the study of AMF effects on the soil carbon cycle, publications between 2001 and 2022 were identified and extracted from the Web of Science Core Collection databases using specific keywords, and a bibliometric analysis conducted on this data using CiteSpace. The number of publications has gradually increased over time, with a notable acceleration from 2017 onward. Most of the authors did not cooperate on a regular basis, and a stable cooperative group had not yet been formed among the core authors. The Chinese Academy of Sciences produced the largest number of publications among research institutions. Specifically, China published the largest number of articles in terms of national contributions, but had less influence in collaborative networks. The most desirable research topics mainly included nitrogen and phosphorus nutrients, root colonization, diversity, organic matter and organic carbon, plant growth, and microbial community. In more recent years, research seemed to focus more on AMF-mediated carbon, nitrogen, phosphorus, nutrient exchange, alterations in microbial community structure, advancements in research techniques and methods, and in-depth studies into related mechanisms. This study serves as a valuable reference for future researchers interested in the effects of AMF on the soil carbon cycle, and provides guidelines for future innovative research.

In order to improve the application of teaching resources and reduce delays in the integration process of multimedia network, a rational resource allocation method for multimedia network teaching reform based on Bayesian partition data mining is proposed. Bayesian partition is used to preprocess the multimedia network teaching resources (MNTR), adjusting the recognition probability of MNTR in each partition based on its attributes. By performing Bayesian quantitative classification using samples of MNTR, the prior probability is adjusted through maximization analysis. The partitioned resources undergo sample data mining to obtain the data category collection of all MNTR. A prediction model is then built to forecast the demand for teaching resources at specific times in the future. MNTR can be rationally allocated based on the prediction results. Experimental results demonstrate that this method reduces delays in MNTR application and improves the accuracy and utilization of teaching resources.

Traditional virtual simulated interaction systems experience data fragmentation during the process of converting from two-dimensional to three-dimensional information, resulting in reduced realism and an inability to meet the teaching requirements of computer courses. Therefore, the integration of augmented reality (AR) technology into the educational environment remains an urgent and unresolved issue. To address the aforementioned issues, this paper investigates the data throughput limitations present in virtual simulation interaction systems. In response to this problem, an application solution utilizing AR technology is proposed, specifically a design concept for a virtual simulation interactive system tailored to computer-related courses. This system achieves its objectives through the collaborative interaction of AR hardware and supplementary software algorithms. The AR hardware is subdivided into framework design and functional hardware design, while the software components encompass AR models, virtual interaction models, and fusion methods. Through testing and comparison of the data throughput of this system with two other virtual simulation interaction systems, it was found that the virtual simulation interactive system optimized using AR technology can effectively enhance data throughput and address the issue of reduced realism in virtual interaction scenes caused by data fragmentation. This design system provides a more realistic and efficient mode of interaction for teaching computer-related courses.

Abstract In this article, by analyzing the content and characteristics of the cost management of the distribution network project, the evaluation index system of the cost management level is formulated first. Gini index in random forest algorithm and analytic hierarchy process are used to analyze the influencing factors of the cost management level of distribution network project from quantitative and qualitative aspects. The results show that design change and on‐site visa management have a great impact on management level, while project advance payment management has a relatively small impact. Based on the above analysis results, this article simplifies the indicators, so as to optimize the random forest model. After verification, the optimized model can improve the efficiency of evaluation, on the other hand, it can effectively improve the accuracy of evaluation, ensure the scientificity of evaluation, and provide reference for improving the cost management of distribution network project.

Summary: Mosaic symptoms are commonly observed in virus-infected plants. However, the underlying mechanism by which viruses cause mosaic symptoms as well as the key regulator(s) involved in this process remain unclear. Here, we investigate maize dwarf mosaic disease caused by sugarcane mosaic virus (SCMV). We find that the manifestation of mosaic symptoms in SCMV-infected maize plants requires light illumination and is correlated with mitochondrial reactive oxidative species (mROS) accumulation. The transcriptomic and metabolomic analyses results together with the genetic and cytopathological evidence indicate that malate and malate circulation pathways play essential roles in promoting mosaic symptom development. Specifically, at the pre-symptomatic infection stage or infection front, SCMV infection elevates the enzymatic activity of pyruvate orthophosphate dikinase by decreasing the phosphorylation of threonine527 under light, resulting in malate overproduction and subsequent mROS accumulation. Our findings indicate that activated malate circulation contributes to the manifestation of light-dependent mosaic symptoms via mROS.

With the development of IoT and 5G technologies, more and more online resources are presented in trendy multimodal data forms over the Internet. Hence, effectively processing multimodal information is significant to the development of various online applications, including e-learning and digital health, to just name a few. However, most AI-driven systems or models can only handle limited forms of information. In this study, we investigate the correlation between natural language processing (NLP) and pattern recognition, trying to apply the mainstream approaches and models used in the computer vision (CV) to the task of NLP. Based on two different Twitter datasets, we propose a convolutional neural network based model to interpret the content of short text with different goals and application backgrounds. The experiments have demonstrated that our proposed model shows fairly competitive performance compared to the mainstream recurrent neural network based NLP models such as bidirectional long short-term memory (Bi-LSTM) and bidirectional gate recurrent unit (Bi-GRU). Moreover, the experimental results also demonstrate that the proposed model can precisely locate the key information in the given text.

Viruses often establish their own infection by altering host metabolism. How viruses co-opt plant metabolism to support their successful infection remains an open question. Here, we used untargeted metabolomics to reveal that lactate accumulates immediately before and after robust sugarcane mosaic virus (SCMV) infection. Induction of lactate-involved anaerobic glycolysis is beneficial to SCMV infection. The enzyme activity and transcriptional levels of lactate dehydrogenase (LDH) were up-regulated by SCMV infection, and LDH is essential for robust SCMV infection. Moreover, LDH relocates in viral replicase complexes (VRCs) by interacting with SCMV-encoded 6K2 protein, a key protein responsible for inducing VRCs. Additionally, lactate could promote SCMV infection by suppressing plant defense responses. Taken together, we have revealed a viral strategy to manipulate host metabolism to support replication compartment but also depress the defense response during the process of infection.

Deep learning (DL), as the core technology of artificial intelligence (AI), has been extensively researched in the past decades. However, practical DL education needs large marked datasets and computing resources, which is generally not easy for students at school. Therefore, due to training datasets and computing resources restrictions, it is still challenging to popularize DL education in colleges and universities. This paper considers solving this problem by collective intelligence from a resource sharing perspective. In DL, dataset marking and model training both require high workforce and computing power, which may implement through a resource sharing mechanism using collective intelligence. As a test, we have designed a DL education scheme based on collective intelligence under the background of artistic creation to collect teaching materials for DL education. Also, we elaborate on the detailed methods of sharing mechanisms in this article and discuss some related problems to verify this shared learning mechanism.

A newly emerged micro learning service offers a flexible formal, informal, or non-formal online learning opportunity to worldwide users with different backgrounds in real-time. With the assist of big data technology and cloud computing service, online learners can access tremendous fine-grained learning resources through micro learning service. However, big data also causes serious information overload during online learning activities. Hence, an intelligent recommender system is required to filter out not-suitable learning resources and pick the one that matches the learner’s learning requirement and academic background. From the perspective of natural language processing (NLP), this study proposed a novel recommender system that utilises machine translation and language modelling. The proposed model aims to overcome the defects of conventional recommender systems and further enhance distinguish ability of the recommender system for different learning resources.

This study investigated the beliefs, attitudes, and motivation of Thai and Chinese undergraduate students in relation to English language learning. Utilizing a mixed-method, explanatory sequential design, it aims to provide a better understanding of the three variables mentioned above. The quantitative data analysis showed disparities between the cohorts, with Chinese undergraduate students often displaying more robust orientations. However, the qualitative data analysis showed that both groups valued immersive learning experiences and technological integration. The research advocates for pedagogical frameworks in English education that align with overarching global paradigms while respecting individual cultural nuances. Such revelations are important for refining teaching practices and curricula to enhance genuine global literacy. It is also recommended that future research efforts diversify their cultural purview and give due consideration to variations in socio-economic contexts.

While the application of manure to improve soil quality has attracted attention, the effect of pig manure application rates on soil acidity remains poorly understood. In this study, we analyzed the changes and correlations between soil acidity, pH buffer capacity (pHBC), soil chemical properties, and crop yields after 8 years of pig manure application at different rates (i.e., 0, 7.5, 15, 30, and 45 Mg ha−1) in a red upland soil (Ultisol). With an increase in the application rates, the crop yields were 0.77–8.85 times higher; the pH was enhanced by 0.4–0.8 units; and the soil organic matter (SOM), pHBC, iron activation (Feo), exchangeable calcium (ExCa), and exchangeable magnesium (ExMg) contents increased by up to 74.1%, 92.7%, 36.7%, 62.2%, and 48.7%, respectively, whereas that of total exchangeable acid (ExAcid) decreased by 17.2–52.9%. The crop yields were positively related to the soil pH but negatively correlated with ExAcid. Redundancy analysis revealed ExAcid and pHBC were more sensitive than pH was to the other chemical indicators. ExAcid was negatively correlated with SOM and ExCa; pHBC was positively correlated with ExMg, TN, SOM, and Feo. In conclusion, the crop yield could be improved by adjusting the soil acidity characteristics, and the application of pig manure reduced the soil acidity, with an optimal application rate of 15 Mg·ha−1.

Road extraction from high-resolution remote sensing images is a critical task in image understanding and analysis, yet it poses significant challenges because of road occlusions caused by vegetation, buildings, and shadows. Deep convolutional neural networks have emerged as the leading approach for road extraction because of their exceptional feature representation capabilities. However, existing methods often yield incomplete and disjointed road extraction results. To address this issue, we propose CR-HR-RoadNet, a novel high-resolution road extraction network that incorporates local and global context reasoning. In this work, we introduce a road-adapted high-resolution network as the feature encoder, effectively preserving intricate details of narrow roads and spatial information. To capture multi-scale local context information and model the interplay between roads and background environments, we integrate multi-scale features with residual learning in a specialized multi-scale feature representation module. Moreover, to enable efficient long-range dependencies between different dimensions and reason the correlation between various road segments, we employ a lightweight coordinate attention module as a global context-aware algorithm. Extensive quantitative and qualitative experiments on three datasets demonstrate that CR-HR-RoadNet achieves superior extraction accuracy across various road datasets, delivering road extraction results with enhanced completeness and continuity. The proposed method holds promise for advancing road extraction in challenging remote sensing scenarios and contributes to the broader field of deep-learning-based image analysis for geospatial applications.

Building polygons plays an important role in urban management. Although leveraging deep learning techniques for building polygon extraction offers advantages, the models heavily rely on a large number of training samples to achieve good generalization performance. In scenarios with small training samples, the models struggle to effectively represent diverse building structures and handle the complexity introduced by the background. A common approach to enhance feature representation is fine-tuning a pre-trained model on a large dataset specific to the task. However, the fine-tuning process tends to overfit the model to the task area samples, leading to the loss of generalization knowledge from the large dataset. To address this challenge and enable the model to inherit the generalization knowledge from the large dataset while learning the characteristics of the task area samples, this paper proposes a knowledge distillation-based framework called Building Polygon Distillation Network (BPDNet). The teacher network of BPDNet is trained on a large building polygon dataset containing diverse building samples. The student network was trained on a small number of available samples from the target area to learn the characteristics of the task area samples. The teacher network provides guidance during the training of the student network, enabling it to learn under the supervision of generalization knowledge. Moreover, to improve the extraction of buildings against the backdrop of a complex urban context, characterized by fuzziness, irregularity, and connectivity issues, BPDNet employs the Dice Loss, which focuses attention on building boundaries. The experimental results demonstrated that BPDNet effectively addresses the problem of limited generalization by integrating the generalization knowledge from the large dataset with the characteristics of the task area samples. It accurately identifies building polygons with diverse structures and alleviates boundary fuzziness and connectivity issues.

Feature selection is a crucial process in machine learning and data mining that identifies the most pertinent and valuable features in a dataset. It enhances the efficacy and precision of predictive models by efficiently reducing the number of features. This reduction improves classification accuracy, lessens the computational burden, and enhances overall performance. This study proposes the improved binary golden jackal optimization (IBGJO) algorithm, an extension of the conventional golden jackal optimization (GJO) algorithm. IBGJO serves as a search strategy for wrapper-based feature selection. It comprises three key factors: a population initialization process with a chaotic tent map (CTM) mechanism that enhances exploitation abilities and guarantees population diversity, an adaptive position update mechanism using cosine similarity to prevent premature convergence, and a binary mechanism well-suited for binary feature selection problems. We evaluated IBGJO on 28 classical datasets from the UC Irvine Machine Learning Repository. The results show that the CTM mechanism and the position update strategy based on cosine similarity proposed in IBGJO can significantly improve the Rate of convergence of the conventional GJO algorithm, and the accuracy is also significantly better than other algorithms. Additionally, we evaluate the effectiveness and performance of the enhanced factors. Our empirical results show that the proposed CTM mechanism and the position update strategy based on cosine similarity can help the conventional GJO algorithm converge faster.

Genetic diversity is associated with invasion dynamics during establishment and expansion stages by affecting the viability and adaptive potential of exotics. There have been many reports on the comparison between the genetic diversity of invasive alien species (IAS) in and out of their native habitats, but the conclusions were usually inconsistent. In this work, a standard meta-analysis of the genetic diversity of 19 invasive plants based on 26 previous studies was carried out to investigate the general trend for the change of IASs’ genetic diversity during their invasion process and its real correlation with the invasion fate. Those 26 studies were screened from a total of 3557 peer-reviewed publications from the ISI Web of Science database during the period of January 2000 to May 2022. Based on the selected studies in this work, a general reduction of IASs’ genetic diversity was found in non-native populations compared to that in native ones, while the difference was not significant. This finding suggested that regardless of the change in genetic diversity, it had no substantial effect on the outcome of the invasion process. Therefore, genetic diversity might not serve as a reliable indicator for risk assessment and prediction of invasion dynamic prediction in the case of IASs.

Semantic segmentation is one of the hot topics in the field of remote sensing image intelligent analysis. Deep convolutional neural network (DCNN) has become a mainstream technology in semantic segmentation due to its powerful semantic feature representation. The emergence of high-resolution remote sensing imagery has provided massive detail information, but difficulties and challenges remain in the “feature representation of fine geo objects” and “feature distinction of easily confusing geo objects.” To this end, this article focuses on the distinguishing features of geo-object details and proposes a novel DCNN-based semantic segmentation. First, the cascaded relation attention module is adopted to determine the relationship among different channels or positions. Then, information connection and error correction are used to capture and fuse the features of geo-object details. The output feature representations are provided by the multiscale feature module. Besides, the proposed model uses the boundary affinity loss to gain accurate and clear geo-object boundary. The experimental results on the Potsdam and Vaihingen datasets demonstrate that the proposed model can achieve excellent segmentation performance on overall accuracy and mean intersection over union. Furthermore, the results of ablation and visualization analyses also verify the feasibility and effectiveness of the proposed method.

In building change detection task, factors such as phenological changes, illumination changes, and registration errors will cause unchanged areas in remote sensing images to have obvious differences in pixels, which will lead to pseudochanges in results. Existing methods focus on the change information of multi-temporal remote sensing images, ignoring the exploration of pseudochange problems. Therefore, feature–output space dual-alignment (FODA) method is proposed to reduce the negative effect of the pseudochange problem by paying attention to the relationship between unchanged areas of multitemporal images. On the one hand, FODA narrows the distance between the features of the unchanged areas in the feature space, increasing its feature extraction ability of pseudo-changed areas. On the other hand, given the spatial context of image scene implicit in the output space, the ability to recognize pseudochanges of the FODA is improved through an adversarial learning procedure. Due to its simplicity and effectiveness, FODA achieves 88.73% and 82.75% F1 scores on the LEVIR-CD dataset and WHU-CD dataset respectively. Compared with state-of-the-art methods, FODA can effectively reduce the problem of pseudo-changes and significantly improve the effect of change detection even only based on a simple backbone model.

A high-quality built environment is important for human health and well-being. Assessing the quality of the urban built environment can provide planners and managers with decision-making for urban renewal to improve resident satisfaction. Many studies evaluate the built environment from the perspective of street scenes, but it is difficult for street-view data to cover every area of the built environment and its update frequency is low, which cannot meet the requirement of built-environment assessment under rapid urban development. Earth-observation data have the advantages of wide coverage, high update frequency, and good availability. This paper proposes an intelligent evaluation method for urban built environments based on scene understanding of high-resolution remote-sensing images. It contributes not only the assessment criteria for the built environment in remote-sensing images from the perspective of visual cognition but also an image-caption dataset applicable to urban-built-environment assessment. The results show that the proposed deep-learning-driven method can provide a feasible paradigm for representing high-resolution remote-sensing image scenes and large-scale urban-built-area assessment.

The sluggish electrochemical kinetics of sulfur species remains a major hurdle for the broad adoption of lithium-sulfur batteries. Here, the authors construct an energy diagram of sulfur species to unveil their reaction pathways and propose a general strategy to accelerate electrochemical reactions.

Antenna arrays are able to improve the directivity performance and reduce the cost of wireless communication systems. However, how to reduce the maximum sidelobe level (SLL) of the beam pattern is a key problem in antenna arrays. In this paper, three kinds of antenna arrays that are linear antenna array (LAA), circular antenna array (CAA) and random antenna array (RAA) are investigated. First, we formulate the SLL suppression optimization problems of LAA, CAA and RAA, respectively. Then, we propose a novel method called improved chicken swarm optimization (ICSO) approach to solve the formulated optimization problems. ICSO introduces four enhanced strategies including the local search factor, weighting factor and global search factor into the update method of conventional chicken swarm optimization (CSO) algorithm, respectively, for achieving better beam pattern optimization results of antenna arrays. Moreover, a variation mechanism is proposed to enhance the population diversity so that further improving the performance of the algorithm. We conduct simulations to evaluate the performance of the proposed ICSO for the maximum SLL suppressions of LAAs, CAAs and RAAs, and the results show that ICSO obtains lower maximum SLLs for different antenna array cases with different numbers of antenna elements compared to several other algorithms.

Feature selection is an effective approach to reduce the number of features of data, which enhances the performance of classification in machine learning. In this paper, we formulate a joint feature selection problem to reduce the number of the selected features while enhancing the accuracy. An improved binary particle swarm optimization (IBPSO) algorithm is proposed to solve the formulated problem. IBPSO introduces a local search factor based on Lévy flight, a global search factor based on weighting inertia coefficient, a population diversity improvement factor based on mutation mechanism and a binary mechanism to improve the performance of conventional PSO and to make it suitable for the binary feature selection problems. Experiments based on 16 classical datasets are selected to test the effectiveness of the proposed IBPSO algorithm, and the results demonstrate that IBPSO has better performance than some other comparison algorithms.

The beam pattern synthesis of antenna arrays is one of the most significant optimization problems in the electromagnetics community, and it affects the performance of wireless communication systems. In this article, we formulate the beam pattern optimization problems, and propose an improved invasive weed optimization with random mutation and Lévy flight (IWORMLF) algorithm to synthesize the beam patterns of linear antenna arrays (LAAs) and circular antenna arrays (CAAs). IWORMLF introduces the random mutation operator and the Lévy flight mechanism to enhance the efficiency and to balance the exploration and exploitation of the algorithm for optimization problems. Several simulations are conducted to evaluate the performance of the proposed approach. First, IWORMLF is tested by a variety of benchmark functions, and the results indicate that it achieves the best performance in most of the functions compared to some other algorithms. Second, IWORMLF is utilized for the maximum sidelobe level (SLL) reduction and the joint maximum SLL and mainlobe beamwidth reductions. The results show that it achieves better performances in terms of the convergence rate, stability, and accuracy compared to some evolutionary algorithms for these optimization cases. Third, the efficiencies of the improved factors are verified. Finally, electromagnetism simulations are conducted to evaluate the performance of IWORMLF for the beam pattern optimizations with considering mutual coupling.

Feature selection is an effective method to eliminate irrelevant, redundant and noisy features, which improves the performance of classification and reduces the computational burden in machine learning. In this paper, an improved binary dragonfly algorithm (IBDA) which extends from the conventional dragonfly algorithm (DA) is proposed as a search strategy to design a wrapper-based feature selection method. First, a novel evolutionary population dynamics (EPD) strategy is introduced in IBDA to enhance the exploitation ability while ensuring population diversity of the algorithm. Second, IBDA proposes a novel crossover operator which establishes connections between the crossover rates and iterations so that making the algorithm can adjust the crossover rates of solutions dynamically, thereby balancing the exploitation and exploration of the algorithm. Finally, a binary mechanism is proposed to make the algorithm suitable for the binary feature selection problems. Simulations are conducted on 27 classical datasets from the UC Irvine Machine Learning Repository, and the results demonstrate that the proposed IBDA has better performance than some other comparison algorithms. Moreover, the effectiveness and performance of the proposed improved factors are evaluated by tests.

Building extraction using deep learning techniques has advantages but relies on a large number of clean labeled samples to train the model. Complex appearance and tilt shots often cause many offsets between building labels and true locations, and these noises have a considerable impact on building extraction. This paper proposes a new knowledge distillation-based building extraction method to reduce the impact of noise on the model and maintain the generalization of the model. The method can maximize the generalizable knowledge of large-scale noisy samples and the accurate supervision of small-scale clean samples. The proposed method comprises two similar teacher and student networks, where the teacher network is trained by large-scale noisy samples and the student network is trained by small-scale clean samples and guided by the knowledge of the teacher network. Experimental results show that the student network can not only alleviate the influence of noise labels but also obtain the capability of building extraction without incorrect labels in the teacher network and improve the performance of building extraction.

Accurate information on grassland above-ground biomass (AGB) is critical to better understanding the carbon cycle and conserve grassland resources. As a climate-sensitive key ecological function area, it is important to accurately estimate the grassland AGB of the Tibetan Plateau. Sentinel-2 (S2) images have advantages in reducing mixed pixels and the scale effect for remote sensing, while the data volume is correspondingly larger. In order to improve the estimation accuracy while reducing the data volume required for AGB estimation and improving the computational efficiency, this study used the Recursive Feature Elimination (RFE) algorithm to find the optimal feature set and compared the performance of the Cubist, Gradient Boosting Regression Tree (GBRT), random forest (RF) and eXtreme Gradient Boosting (XGBoost) algorithms for estimating AGB. In this study, ten S2 bands, ten S2-derived vegetation indexes, 218 pieces of AGB field survey data, four types of meteorological data and three types of topographic data were used as the alternative input features for the AGB estimation model. The impurity and permutation importance were used as the feature importance calculation method input to the RFE, and the Cubist, GBRT, RF and XGBoost algorithms were used to construct the AGB estimation models. The results showed that the RF algorithm based on the monthly average temperature (T), elevation, Normalized Difference Phenology Index (NDPI), Normalized Difference Infrared Index (NDII) and Palmer Drought Severity Index (PDSI) performed best (R2 = 0.8838, RMSE = 35.05 g/m2, LCCC = 2.44, RPPD = 0.91). The above findings suggest that the RF model based on the features related to temperature, altitude, humidity and leaf water content is beneficial to estimate the grassland AGB on the Tibetan Plateau.

Excessive fertilization is often applied to produce rice. To reduce nitrogen loss and improve nitrogen use efficiency (NUE), we studied the effects of application depth (surface application, 5 and 10 cm) and shape of nitrogen fertilizers (row application and deep application of large granular fertilizer) on rice growth, soil N distribution and ammonia volatilization. The results showed that grain yield, shoot biomass and total dry biomass of the treatment with N in large granular fertilizer applied at 10 cm depth were significantly higher than those of all other treatments. Moreover, compared with the surface application, the N recovery efficiency and the N agronomic efficiency of deep application treatments were enhanced by 18.1–52.3% and 35.6–95.6%, respectively. Deep application significantly increased NH4+-N concentration at their fertilization points. During the growth season, N in large granular fertilizer treatments (mixed with clay to form an unusually large pellet of 1.0–1.5 cm in diameter) distributed closer to the roots, while N in other treatments, including row application treatments, was more widely distributed. Compared with the surface application, deep application significantly reduced NH3 volatilization and NH4+-N concentration in surface water by 58.7–64.8% and 26.0–72.5%, respectively. Furthermore, the NH3 volatilization from large granular treatment was 7.6–11.0% lower than that in the row application. In conclusion, applying N in large granular fertilizer at 10 cm depth reduces ammonia volatilization, and improves rice growth and grain yield, indicating improved NUE and lowered environmental risks.

Service recommendation is key to improving users’ online experience. The development of the Internet has accelerated the creation of many services, and whether users can obtain good experiences among the massive number of services mainly depends on the quality of service recommendation. It is commonly believed that deep learning has excellent nonlinear fitting ability in capturing the complex interactions between users and items. The advantage in learning intricacy relationships enables deep learning to become an important technology for present service recommendation. Recently, it is noticed that linear models can perform almost as well as the state-of-the-art deep learning models, suggesting that capturing linear relationships between users and items is also very important for recommender systems. Therefore, numerous deep learning systems combined with linear models have been proposed. However, existing models are incapable of considering the size of the embedding. When the embedding dimension is too large, it leads to overfitting and thus influences the model’s ability to capture linear relationships. In this paper, a neural network based on two-layer matrix factorization and multi-layer perceptron—Two-layer Matrix factorization and Multi-layer perceptron Neural Network (TMMNN)—is proposed. To solve the problem of overfitting caused by an oversized embedding dimension, multi-size embedding technology has been integrated into the model. Matrix factorization and the multi-layer perceptron are placed in the upper and lower layers respectively, and they both receive embedding vectors dynamically adjusted for dimensions. In the upper layer, the matrix factorization is responsible for receiving the embedding of users and items, capturing linear relationships, and then yielding the generated new vectors as input to the multi-layer perceptron in the lower layer. Compared to other previously proposed models, the experimental results on the standard datasets MovieLens 20M and MovieLens Latest show that the TMMNN model is evidently better in terms of prediction accuracy.

The power pattern synthesis, as well as the sidelobe level (SLL) suppression of the planar antenna arrays for the wireless power transmission is investigated. An improved chicken swarm optimization (ICSO) algorithm is proposed for reducing the maximum SLL of the power pattern. ICSO introduces the global search, the local search, and the duplicate solution removing operators to enhance the accuracy and the convergence rate of the conventional CSO algorithm, thereby making it suitable for the power pattern optimizations. Simulations are conducted to verify the performance of the proposed algorithm, and the results show that ICSO obtains the lowest maximum SLL compared with other algorithms. Moreover, the power pattern performances of different algorithms are tested by the electromagnetic simulations, and the results show that the proposed ICSO has the best optimization ability.

Conventional collaborative beamforming (CB) with virtual node antenna array often results in high maximum sidelobe level (SLL) due to the unexpected positions of the nodes. In this paper, a sidelobe suppression approach (JSSA) to suppress the maximum SLL by jointly optimizing the locations and amplitude weights of the nodes is proposed. JSSA organizes the node positions according to the concentric circular antenna array for location optimization. Then, a novel algorithm called variation particle chicken swarm optimization (VPCSO) is proposed to further optimize the amplitude weights of the selected CB nodes. The proposed VPCSO introduces the variation mechanism and location learning mechanism to enhance the performance of the conventional chicken swarm optimization algorithm. Simulations are conducted and the results show that the proposed location optimization approach is effective, and the maximum SLLs of beam patterns obtained by VPCSO are lower than that of other algorithms. Moreover, the energy consumption can be saved by VPCSO. In addition, electromagnetic (EM) simulations are conducted to verify the performance of the proposed JSSA in EM environments.

China’s urbanization has dramatically accelerated in recent decades. Land for urban build-up has changed not only in large cities but also in small counties. Land cover mapping is one of the fundamental tasks in the field of remote sensing and has received great attention. However, most current mapping requires a significant manual effort for labeling or classification. It is of great practical value to use the existing low-resolution label data for the classification of higher resolution images. In this regard, this work proposes a method based on noise-label learning for fine-grained mapping of urban build-up land in a county in central China. Specifically, this work produces a build-up land map with a resolution of 10 m based on a land cover map with a resolution of 30 m. Experimental results show that the accuracy of the results is improved by 5.5% compared with that of the baseline method. This notion indicates that the time required to produce a fine land cover map can be significantly reduced using existing coarse-grained data.

Dispersing liquid metal droplets in a soft elastomer matrix yields a composite, called liquid metal embedded elastomer (LMEE), with less modified stiffness but noticeably increased relative permittivity. Stretching and applying voltage through the thickness of the elastomer composite directly, it is vulnerable to the loss of being an insulator due to the conductive pathway formed during prestretching. Here, we describe a very simple sandwich structure that enables electromechanical actuation of LMEE. It consists of two very high bonding tapes as skins and a layer of LMEE as the core, making the sandwich all polymeric and thus stretchable. The electromechanical performance of a LMEE made of Ecoflex and Galinstan at various mass ratios was systematically measured. For a typical circular actuation made of the sandwich structure, a 90% areal actuation strain was achieved. Our efforts pave the way for various applications of LMEE.

The beam pattern of the virtual node antenna array (VNAA) in wireless sensor networks has a high maximum sidelobe level (SLL), thereby causing communication interference because of the uncontrollable node positions. A sidelobe and energy optimization array node selection (SEOANS) algorithm is proposed for optimally selecting the sensor nodes to form a VNAA that optimizes the beam pattern of VNAA and reduces the average energy consumption of nodes. SEOANS uses a calculation method to determine the optimal number of array nodes, proposes a node location selection optimization method based on concentric circular ring array and a novel swarm intelligence optimization algorithm called cuckoo search chicken swarm optimization (CSCSO) to optimize the excitation current of each array node. CSCSO uses chaos theory, introduces the inertia weight Lévy flight, and adopts the grade mechanism in chicken swarm optimization to improve the performance of the cuckoo search algorithm. In addition, the scheduling and fault tolerance mechanisms are designed and implemented in SEOANS. Simulation results show that the node position selection optimization method and the excitation current optimization based on CSCSO can effectively reduce the maximum SLL. Furthermore, compared with traditional clustering routing algorithms, SEOANS has advantages in the communication delay and average energy consumption of nodes, thus effectively improving network lifetime.

Flying safely in complex urban environments is a challenge for unmanned aerial vehicles because path planning in urban environments with many narrow passages and few dynamic flight obstacles is difficult. The path planning problem is decomposed into global path planning and local path adjustment in this paper. First, a branch-selected rapidly-exploring random tree (BS-RRT) algorithm is proposed to solve the global path planning problem in environments with narrow passages. A cyclic pruning algorithm is proposed to shorten the length of the planned path. Second, the GM(1,1) model is improved with optimized background value named RMGM(1,1) to predict the flight path of dynamic obstacles. Herein, the local path adjustment is made by analyzing the prediction results. BS-RRT demonstrated a faster convergence speed and higher stability in narrow passage environments when compared with RRT, RRT-Connect, P-RRT, 1-0 Bg-RRT, and RRT*. In addition, the path planned by BS-RRT through the use of the cyclic pruning algorithm was the shortest. The prediction error of RMGM(1,1) was compared with those of ECGM(1,1), PCGM(1,1), GM(1,1), MGM(1,1), and GDF. The trajectory predicted by RMGM(1,1) was closer to the actual trajectory. Finally, we use the two methods to realize path planning in urban environments.

Soil microbe and soil enzyme activities are closely related to the fertilizer inputs. Our objective was to explore the dynamic change of soil microbe and soil enzyme activities in paddy field during early and late rice (Oryza sativa L.) main growth stages with different long-term fertilizer managements in the double cropping rice system. we analyzed the soil microbe and soil enzyme activities with mineral fertilizer alone (MF), rice residues and mineral fertilizer (RF), 30% organic matter and 70% mineral fertilizer (LOM), 60% organic matter and 40% mineral fertilizer (HOM), and without fertilizer (CK) basis on long-term fertilizer experiment. The results showed that the soil enzyme activities were increased by application of mineral fertilizer along with manure or rice residues at the main growth stages of early and late rice. The soil urease activities for different fertilizer managements was RF>HOM>LOM>MF>CK at the main growth stages of early and late rice. The soil catalase activities with HOM, LOM treatments was highest, and was significantly higher than that of RF, MF, CK treatments at the main growth stages of rice. The soil invertase and dehydrogenase activities with HOM treatment were highest, and were significantly higher than that of CK at the main growth stages of rice. And the soil cellobiohyrolase activities with RF treatment were highest, and were significantly higher than that of MF, CK treatments at the main growth stages of rice. Meanwhile, the results indicated that different fertilization managements were significantly affected some physiological function soil microbial quantity. The amount of nitrifying and denitrification bacteria in soil for different fertilizer managements was HOM>LOM>RF>MF>CK at the main growth stages of early and late rice. Meanwhile, the amount of ammonifiers and cellulose”decomposing bacteria in soil for different fertilizer managements was MF>RF>LOM>HOM>CK and LOM>RF>MF>HOM>CK at the main growth stages of rice, respectively. The amount of azotobacteria in soil with RF treatment was highest, and was significantly higher than that of MF, CK treatments at the main growth stages of rice. As a result, the soil microbe and soil enzyme activities were increased by applied of fertilizer practices, the effect of mineral fertilizer combined with manure or rice residues is better than that of only mineral fertilizer.

Synechococcus is one of the most important contributors to global primary productivity, and ocean warming is predicted to increase abundance and distribution of Synechococcus in the ocean. Here, we investigated molecular response of an oceanic Synechococcus strain WH8102 grown in two nitrogen sources (nitrate and urea) under present (25°C) and predicted future (28°C) temperature conditions using an isobaric tag (IBT)-based quantitative proteomic approach. Rising temperature decreased growth rate, contents of chlorophyll a, protein and sugar in the nitrate-grown cells, but only decreased protein content and significantly increased zeaxanthin content of the urea-grown cells. Expressions of CsoS2 protein involved in carboxysome formation and ribosomal subunits in both nitrate- and urea-grown cells were significantly decreased in rising temperature, whereas carbohydrate selective porin and sucrose-phosphate synthase (SPS) were remarkably up-regulated, and carbohydrate degradation associated proteins, i.e., glycogen phosphorylase kinase, fructokinase and glucose-6-phosphate dehydrogenase, were down-regulated in the urea-grown cells. Rising temperature also increased expressions of three redox-sensitive enzymes (peroxiredoxin, thioredoxin, and CP12) in both nitrate- and urea-grown cells. Our results indicated that rising temperature did not enhance cell growth of Synechococcus; on the contrary, it impaired cell functions, and this might influence cell abundance and distribution of Synechococcus in a future ocean.

Understanding the connectivity between surface water and groundwater is key to sound geo-hazard prevention and mitigation in a waterscape such as the Jiuzhaigou Natural World Heritage Site in the southeast Tibetan Plateau, China. In this study, we used environmental isotope tracers (2H, 18O 3H, and 222Rn) to constrain a water cycle model including confirming hydrological pathways, connectivity, and water source identification in the Jiuzhaigou catchments. We established the local meteoric water line (LMWL) based on the weekly precipitation isotope sampling of a precipitation station. We systematically collected water samples from various water bodies in the study area to design the local water cycle model. The regional water level and discharge changes at one month after the earthquake indicated that there was a hydraulic connection underground across the local water divide between the Rize (RZ) river in the west and Zechawa (ZCW) lake in the east by the δ18O and δ2H measurements. We employed an end-member mixing model to identify and quantify Jiuzhaigou runoff-generating sources and their contributions, and we found that the average contributions of precipitation and groundwater to the surface runoff in the catchments are about 30% and 70%, respectively. The two branches of the Shuzheng (SZ) trunk were recharged by 62 ± 19% from the ZCW lake and 38 ± 19% from the RZ river, which was consistent with the fractions calculated by the actual discharge volume. 222Rn mass balance analyses were employed to estimate the water exchange between groundwater and river, which further confirmed this estimate. 222Rn concentrations and 3H contents showed that the groundwater had a short residence time and it was moderate precipitation, thought the contribution of groundwater to the river was 70%, according to the different tracers. A three-dimensional conceptual model of the water cycle that integrated the regional hydrological and geological conditions was established for the catchments.

The lack of pixel-level labeling limits the practicality of deep learning-based building semantic segmentation. Weakly supervised semantic segmentation based on image-level labeling results in incomplete object regions and missing boundary information. This paper proposes a weakly supervised semantic segmentation method for building detection. The proposed method takes the image-level label as supervision information in a classification network that combines superpixel pooling and multi-scale feature fusion structures. The main advantage of the proposed strategy is its ability to improve the intactness and boundary accuracy of a detected building. Our method achieves impressive results on two 2D semantic labeling datasets, which outperform some competing weakly supervised methods and are close to the result of the fully supervised method.

The karst lakes in Jiuzhaigou Nature Reserve are an integral part of the karst lake landscape, yet research on the formation mechanism for the color of the blue-green lakes in Jiuzhaigou is insufficient. With the help of hyperspectral instruments, coupled with hydro-chemical analysis, this paper elaborates on the unique color characteristics of the Jiuzhaigou karst lakes, delves into the color formation mechanism of the lakes, establishes a regression equation for the color of the lakes as well as the water quality parameters, and sheds light upon the causes for the color distinction between the karst lakes in Jiuzhaigou and the plateau freshwater lakes. The experiment shows that the Jiuzhaigou karst lakes are primarily blue and green, while the proportion of short-wavelength light in the normalized water-leaving radiance and the total incident irradiance of lake water is higher. Based on the redundancy analysis and the correlation analysis, travertine deposition is the core link in the color formation of the blue karst lakes in Jiuzhaigou, while the selective reflection and scattering of the suspended calcium carbonate particulate matters towards visible light represents the optical foundation for the formation. In addition, physical factors such as depth and transparency, changes to the water quality parameters that affect the travertine deposition rate, and the eutrophication process will all exert significant influence over the formation. By building on water-leaving radiance, this paper quantifies the lake color with the tristimulus values (R, G, B) via colorimetrical methods, which features solid goodness of fit with the linear regression equation established based on the water quality parameters. The principal component analysis and colorimetrical analysis show that the color of the karst lakes in Jiuzhaigou varies substantially from that of the plateau freshwater lakes, which mainly results from the difference in the water quality. Research conducted in this paper on the color formation mechanism of the distinct blue karst lakes in Jiuzhaigou illuminates the formation and maintenance mechanism of the plateau karst lakes, which is conducive to better understanding towards the relationship between the water quality and colors of the karst lakes, and provides scientific proof for the establishment of the water quality assessment indicator system based on the colors of the karst lakes.

ABSTRACT Soil microbial biomass plays a significant role in soils, and it is often used as an early indicator of change in soil quality. Soil microbial biomass is affected by different fertilization management practices. Therefore, the impact of different long-term fertilization management practices on the soil organic carbon (SOC) content, soil microbial biomass carbon (SMBC), and soil microbial biomass nitrogen (SMBN), as well as the soil microbial quotient (SMQ) in the tilled layer (0.00-0.20 m) were studied in the present paper, together with grain yield, in a double-cropping rice (Oryza sativa L.) system. The experiment in NingXiang county of Hunan Province, China, begin in 1986, and the experiment included five fertilization treatments: without fertilizer input (CK), mineral fertilizer alone (MF), rice straw residues and mineral fertilizer (RF), 30 % organic matter and 70% mineral fertilizer (LOM), and 60% organic matter and 40 % mineral fertilizer (HOM). The results showed that there is no significant difference in effect on SOC, SMBC, and SMBN contents and on the SMQ in the paddy field with MF treatment compared with the CK treatment at the main growth stages of early and late rice. The SOC, SMBC, SMBN contents, and the SMQ in the paddy field were highest in the LOM and HOM treatments, followed by the RF treatment, at the main growth stages of early and late rice. The results indicated that grain yields of early and late rice with the LOM, HOM, and RF treatments were higher than the yields under the MF and CK treatments. As a result, combined application of organic matter or rice straw residues with mineral fertilizer is a practice available for increasing SOC and microbial biomass contents in double-cropping rice paddy soils.

The problems of synthesizing the beam patterns of the linear antenna array (LAA) and the circular antenna array (CAA) are addressed. First, an optimization problem is formulated for reducing the maximum sidelobe level (SLL) of the beam patterns. Then, the formulated problem is solved by using the invasive weed optimization (IWO) algorithm. Various simulations are performed to evaluate the effectiveness of the IWO algorithm for the synthesis of the beam patterns of the LAA and the CAA. The results show that IWO has a better performance in terms of the accuracy, the convergence rate, and the stability compared with other algorithms for the SLL reductions. Moreover, the electromagnetic simulation results also show that IWO achieves the best performance for the beam pattern synthesis of the antenna arrays in practical conditions.

The network planning is a key factor that directly affects the performance of the wireless networks. Distributed antenna system (DAS) is an effective strategy for the network planning. This paper investigates the antenna deployment in a DAS for the high-speed railway communication networks and formulates an optimization problem which is NP-hard for achieving the optimal deployment of the antennas in the DAS. To solve this problem, a scheme based on an improved cuckoo search based on dimension cells (ICSDC) algorithm is proposed. ICSDC introduces the dimension cell mechanism to avoid the internal dimension interferences in order to improve the performance of the algorithm. Simulation results show that the proposed ICSDC-based scheme obtains a lower network cost compared with the uniform network planning method. Moreover, ICSDC algorithm has better performance in terms of the convergence rate and accuracy compared with the conventional cuckoo search algorithm, the particle swarm optimization, and the firefly algorithm.

For the target problem of a directional wireless sensor network, the greedy algorithm can easily fall into the local optimal solution, whereas the genetic algorithm must forecast the lifetime of the upper bound of the network. We propose a novel multi-objective coverage optimization memetic algorithm that encodes the solutions as chromosomes and simulates the biological evolution process in search for a favourable solution to address the aforementioned problems. Experimental results show that the proposed algorithm can prolong the network lifetime more effectively than similar heuristic algorithms in other studies.