Search

This study investigated the navigation problem for cellular-connected unmanned aerial vehicles (UAVs), particularly in highly dynamic urban environments. To address this problem, the UAV is required not only to evade high-speed obstacles in the airspace but also to avoid the coverage holes of cellular base stations (BS). Moreover, the UAV needs to reach the destination to complete the navigation task. Hence, it is imperative to design the trade-off in action selections between collision evasion and destination-approaching scenarios, while also considering the expected communication outage duration as a crucial reference. To overcome this multiobjective optimization challenge, we propose a deep reinforcement learning (DRL)-based algorithm aimed at enabling the UAV to acquire an optimal decision-making policy. Specifically, we formulated the navigation problem as a Markov decision process (MDP) and developed a layered recurrent soft actor–critic (RSAC)-based DRL framework, stimulating the UAV to resolve two fundamental subtasks of UAV navigation. Furthermore, we develop a multilayer perception (MLP)-based integrated evaluation network to select a particular action from the two subsolutions, satisfying the demands for the entire navigation problem. The layered architecture simplifies the navigation problem, thereby enhancing the convergence speed of the proposed algorithm. Numerical results indicate that the layered-RSAC-based UAV can autonomously perform scheduled navigation tasks in our designed simulated urban environments with superior effectiveness. [ABSTRACT FROM AUTHOR]

Robots assist emergency responders by collecting critical information remotely. Deploying multiple cooperative unmanned ground vehicles (UGVs) for a response can reduce the response time, improve situational awareness, and minimize costs. Reliable communication is critical for multiple UGVs for environmental response because multiple robots need to share information for cooperative navigation and data collection. In this work, we investigate a control policy for optimal communication among multiple UGVs and base stations (BSs). A multi-agent deep deterministic policy gradient (MADDPG) algorithm is proposed to update the control policy for the maximum signal-to-interference ratio. The UGVs communicate with both the fixed BSs and a mobile BS. The proposed control policy can navigate the UGVs and mobile BS to optimize communication and signal strength. Finally, a genetic algorithm (GA) is proposed to optimize the hyperparameters of the MADDPG-based training. Simulation results demonstrate the computational efficiency and robustness of the GA-based MADDPG algorithm for the control of multiple UGVs. [ABSTRACT FROM AUTHOR]

In the field of photovoltaic (PV) power prediction, long-term forecasting, which is more challenging than short-term forecasting, can provide more comprehensive and forward-looking guidance. Currently, significant achievements have been made in the field of short-term forecasting for PV power, but inadequate attention has been paid to long-term forecasting. Additionally, multivariate global forecasting across multiple sites and the limited historical time series data available further increase the difficulty of prediction. To address these challenges, we propose a variable–adaptive channel-independent architecture (VACI) and design a deep tree-structured multi-scale gated component named DTM block for this architecture. Subsequently, we construct a specific forecasting model called DTMGNet. Unlike channel-independent modeling and channel-dependent modeling, the VACI integrates the advantages of both and emphasizes the diversity of training data and the model's adaptability to different variables across channels. Finally, the effectiveness of the DTM block is empirically validated using the real-world solar energy benchmark dataset. And on this dataset, the multivariate long-term forecasting performance of DTMGNet achieved state-of-the-art (SOTA) levels, particularly making significant breakthroughs in the 720-step ultra-long forecasting window, where it reduced the MSE metric below 0.2 for the first time (from 0.215 to 0.199), representing a reduction of 7.44%. [ABSTRACT FROM AUTHOR]

Temporal knowledge graphs (TKGs) are used for dynamically modeling facts in the temporal dimension, and are widely used in various fields. However, existing reasoning models often fail to consider the similarity features between entity relationships and static attributes, making it difficult for them to effectively handle these temporal attributes. Therefore, these models have limitations in dealing with previously invisible entities that appear over time and the implicit associations of static attributes between entities. To address this issue, we propose a temporal knowledge graph reasoning model based on Entity Relationship Similarity Perception, known as ERSP. This model employs the similarity measurement method to capture the similarity features of entity relationships and static attributes, and then fuses these features to generate structural representations. Finally, we provide a decoder with entity relationship representation, static attribute representation, and structural representation information to form a quadruple. Experiments conducted on five common benchmark datasets show that ERSP surpasses the majority of TKG reasoning methods. [ABSTRACT FROM AUTHOR]

This paper investigates the state estimation and event-triggered control for positive nonlinear multi-agent systems. Firstly, a proportional–integral observer is established to estimate the states of the considered nonlinear positive multi-agent systems based on the matrix decomposition method. Then, a dynamic event-triggered mechanism is constructed, and a control protocol is proposed based on the proportional–integral observer and event-triggered mechanism. By combining linear programming with linear co-positive Lyapunov functions, the considered multi-agent systems are guaranteed to be positive and achieve consensus. Moreover, by introducing three new variables and a finite vector, the final convergence point can be changed based on the given vector. Finally, two illustrative examples demonstrate the validity of the proposed theoretical results. [ABSTRACT FROM AUTHOR]

In response to the problems of random selection of primary nodes, high communication complexity, and low consensus efficiency in the current consensus mechanism for intellectual property transactions, a Practical Byzantine Fault Tolerance (PBFT) consensus algorithm based on the Affinity-Propagation (AP) clustering algorithm, termed AP-PBFT, is proposed. Firstly, the election strategy of the leader node is constructed based on the reputation mechanism; the reward and punishment mechanism is designed to achieve the dynamic adjustment of the reputation value of the nodes in the PBFT consensus process, and the number of votes among the nodes is introduced to determine the node's reputation value in collaboration with the reward and punishment mechanism to guarantee the precise ordering of the nodes. Secondly, nodes with high reputation values are selected as cluster centers to run the AP clustering algorithm, and clustering groups of knowledge property transaction nodes are constructed based on responsibility and availability. Finally, the three-stage consensus process of the PBFT consensus algorithm is optimized, and the consensus task is decomposed into two layers: the intra-consensus group and the inter-leader node group, reducing the communication complexity of transaction data in the blockchain. Experimental findings indicate a significant performance improvement of the algorithm over the PBFT consensus algorithm in communication complexity, throughput, and consensus efficiency in the simulation environment of multiple types of transactions in intellectual property transactions, including different types of large-scale transaction scenarios, such as purchases, sales, licenses, and transfers. [ABSTRACT FROM AUTHOR]

In response to the shortcomings of the consensus algorithm for intellectual property transactions, such as high communication overhead, random primary node selection, and prolonged consensus time, a Practical Byzantine Fault Tolerance (PBFT) improvement algorithm based on Divisive Analysis (DIANA) D-PBFT algorithm is proposed. Firstly, the algorithm adopts the hierarchical clustering mechanism of DIANA to cluster nodes based on similarity, enhancing node partition accuracy and reducing the number of participating consensus nodes. Secondly, it designs a reward and punishment system based on node ranking, to achieve consistency between node status and permissions, timely evaluation, and feedback on node behaviours, thereby enhancing node enthusiasm. Then, the election method of the primary node is improved by constructing proxy and alternate nodes and adopting a majority voting strategy to achieve the selection and reliability of the primary node. Finally, the consistency protocol is optimised to perform consensus once within the cluster and once between all primary nodes, to ensure the accuracy of the consensus results. Experimental results demonstrate that the D-PBFT algorithm shows a better performance, in terms of communication complexity, throughput, and latency. [ABSTRACT FROM AUTHOR]

Sonar imaging technology is widely used in the field of marine and underwater monitoring because sound waves can be transmitted in elastic media, such as the atmosphere and seawater, without much interference. In underwater object detection, due to the unique characteristics of the monitored sonar image, and since the target in an image is often accompanied by its own shadow, we can use the relative relationship between the shadow and the target for detection. To make use of shadow-information-aided detection and realize accurate real-time detection in sonar images, we put forward a network based on a lightweight module. By using the attention mechanism with a global receptive field, the network can make the target pay attention to the shadow information in the global environment, and because of its exquisite design, the computational time of the network is greatly reduced. Specifically, we design a ShuffleBlock model adapted to Hourglass to make the backbone network lighter. The concept of CNN dimension reduction is applied to MHSA to make it more efficient while paying attention to global features. Finally, CenterNet's unreasonable distribution method of positive and negative samples is improved. Simulation experiments were carried out using the proposed sonar object detection dataset. The experimental results further verify that our improved model has obvious advantages over many existing conventional deep learning models. Moreover, the real-time monitoring performance of our proposed model is more conducive to the implementation in the field of ocean monitoring. [ABSTRACT FROM AUTHOR]

Based on the practical Byzantine fault tolerance algorithm (PBFT), a grouped multilayer PBFT consensus algorithm (GM-PBFT) is proposed to be applied to digital asset transactions in view of the problems with excessive communication complexity and low consensus efficiency found in the current consensus mechanism for digital asset transactions. Firstly, the transaction nodes are grouped by type, and each group can handle different types of consensus requests at the same time, which improves the consensus efficiency as well as the accuracy of digital asset transactions. Second, the group develops techniques like validation, auditing, and re-election to enhance Byzantine fault tolerance by thwarting malicious node attacks. This supervisory mechanism is implemented through the Raft consensus algorithm. Finally, the consensus is stratified for the nodes in the group, and the consensus nodes in the upper layer recursively send consensus requests to the lower layer until the consensus request reaches the end layer to ensure the consistency of the block ledger in the group. Based on the results of the experiment, the approach may significantly outperform the PBFT consensus algorithm when it comes to accuracy, efficiency, and preserving the security and reliability of transactions in large-scale network node digital transaction situations. [ABSTRACT FROM AUTHOR]

Chronic diseases are increasingly major threats to older persons, seriously affecting their physical health and well-being. Hospitals have accumulated a wealth of health-related data, including patients' test reports, treatment histories, and diagnostic records, to better understand patients' health, safety, and disease progression. Extracting relevant information from this data enables physicians to provide personalized patient-treatment recommendations. While collaborative filtering techniques and classical algorithms such as naive Bayes, logistic regression, and decision trees have had notable success in health-recommendation systems, most current systems primarily inform users of their likely preferences without providing explanations. This paper proposes an approach of deep learning with a local interpretable model–agnostic explanations (LIME)-based interpretable recommendation system to solve this problem. Specifically, we apply the proposed approach to two chronic diseases common in older adults: heart disease and diabetes. After data preprocessing, we use six deep-learning algorithms to form interpretations. In the heart-disease data set, the actual model recommendation of multi-layer perceptron and gradient-boosting algorithm differs from the local model's recommendation of LIME, which can be used as its approximate prediction. From the feature importance of these two algorithms, it can be seen that the CholCheck, GenHith, and HighBP features are the most important for predicting heart disease. In the diabetes data set, the actual model predictions of the multi-layer perceptron and logistic-regression algorithm were little different from the local model's prediction of LIME, which can be used as its approximate recommendation. Moreover, from the feature importance of the two algorithms, it can be seen that the three features of glucose, BMI, and age were the most important for predicting heart disease. Next, LIME is used to determine the importance of each feature that affected the results of the calculated model. Subsequently, we present the contribution coefficients of these features to the final recommendation. By analyzing the impact of different patient characteristics on the recommendations, our proposed system elucidates the underlying reasons behind these recommendations and enhances patient trust. This approach has important implications for medical recommendation systems and encourages informed decision-making in healthcare. [ABSTRACT FROM AUTHOR]

Extractive text summarization selects the most important sentences from a document, preserves their original meaning, and produces an objective and fact-based summary. It is faster and less computationally intensive than abstract summarization techniques. Learning cross-sentence relationships is crucial for extractive text summarization. However, most of the language models currently in use process text data sequentially, which makes it difficult to capture such inter-sentence relations, especially in long documents. This paper proposes an extractive summarization model based on the graph neural network (GNN) to address this problem. The model effectively represents cross-sentence relationships using a graph-structured document representation. In addition to sentence nodes, we introduce two nodes with different granularity in the graph structure, words and topics, which bring different levels of semantic information. The node representations are updated by the graph attention network (GAT). The final summary is obtained using the binary classification of the sentence nodes. Our text summarization method was demonstrated to be highly effective, as supported by the results of our experiments on the CNN/DM and NYT datasets. To be specific, our approach outperformed baseline models of the same type in terms of ROUGE scores on both datasets, indicating the potential of our proposed model for enhancing text summarization tasks. [ABSTRACT FROM AUTHOR]

The existing models for NL2SQL tasks are mainly oriented toward English text and cannot solve the problems of column name reuse in Chinese text data, description in natural language query, and inconsistent representation of data stored in the database. To address this problem, this paper proposes a Chinese cross-domain NL2SQL model based on a heterogeneous graph and relative position attention mechanism. This model introduces relational structure information defined by the expert to construct initial heterogeneous graphs for database schemas and natural language questions. The heterogeneous graph is pruned based on natural language questions, and the multi-head relative position attention mechanism is used to encode the database schema and natural language questions. The target SQL statement is generated using a tree-structured decoder with predefined SQL syntax. Experimental results on the CSpider dataset demonstrate that our model better aligns database schema with natural language questions and understands the semantic information in natural language queries, effectively improving the matching accuracy of Chinese multi-table SQL statement generation. [ABSTRACT FROM AUTHOR]

As chatbots become more popular, multi-intent spoken language understanding (SLU) has received unprecedented attention. Multi-intent SLU, which primarily comprises the two subtasks of multiple intent detection (ID) and slot filling (SF), has the potential for widespread implementation. The two primary issues with the current approaches are as follows: (1) They cannot solve the problem of slot nesting; (2) The performance and inference rate of the model are not high enough. To address these issues, we suggest a multi-intent joint model based on global pointers to handle nested and non-nested slots. Firstly, we constructed a multi-dimensional type-slot label interaction network (MTLN) for subsequent intent decoding to enhance the implicit correlation between intents and slots, which allows for more adequate information about each other. Secondly, the global pointer network (GP) was introduced, which not only deals with nested and non-nested slots and slot incoherence but also has a faster inference rate and better performance than the baseline model. On two multi-intent datasets, the proposed model achieves state-of-the-art results on MixATIS with 1.6% improvement of intent Acc, 0.1% improvement of slot F1 values, 3.1% improvement of sentence Acc values, and 1.2%, 1.1% and 4.5% performance improvements on MixSNIPS, respectively. Meanwhile, the inference rate is also improved. [ABSTRACT FROM AUTHOR]

Task-oriented dialogue systems depend on dialogue state tracking to keep track of the intentions of users in the course of conversations. Although recent models in dialogue state tracking exhibit good performance, the errors in predicting the value of each slot at the current dialogue turn of these models are easily carried over to the next turn, and unlikely to be revised in the next turn, resulting in error propagation. In this paper, we propose a revisable state prediction for dialogue state tracking, which constructs a two-stage slot value prediction process composed of an original prediction and a revising prediction. The original prediction process jointly models the previous dialogue state and dialogue context to predict the original dialogue state of the current dialogue turn. Then, in order to avoid the errors existing in the original dialogue state continuing to the next dialogue turn, a revising prediction process utilizes the dialogue context to revise errors, alleviating the error propagation. Experiments are conducted on MultiWOZ 2.0, MultiWOZ 2.1, and MultiWOZ 2.4 and results indicate that our model outperforms previous state-of-the-art works, achieving new state-of-the-art performances with 56.35, 58.09, and 75.65% joint goal accuracy, respectively, which has a significant improvement (2.15, 1.73, and 2.03%) over the previous best results. [ABSTRACT FROM AUTHOR]

Introduction: Computed tomography (CT) is a non-invasive examination tool that is widely used in medicine. In this study, we explored its value in visualizing and quantifying coconut. Materials and methods: Twelve coconuts were scanned using CT for three months. Axial CT images of the coconuts were obtained using a dual-source CT scanner. In postprocessing process, various three-dimensional models were created by volume rendering (VR), and the plane sections of different angles were obtained through multiplanar reformation (MPR). The morphological parameters and the CT values of the exocarp, mesocarp, endocarp, embryo, bud, solid endosperm, liquid endosperm, and coconut apple were measured. The analysis of variances was used for temporal repeated measures and linear and non-linear regressions were used to analyze the relationship between the data. Results: The MPR images and VR models provide excellent visualization of the different structures of the coconut. The statistical results showed that the weight of coconut and liquid endosperm volume decreased significantly during the three months, while the CT value of coconut apple decreased slightly. We observed a complete germination of a coconut, its data showed a significant negative correlation between the CT value of the bud and the liquid endosperm volume (y = −2.6955x + 244.91; R2 = 0.9859), and a strong positive correlation between the height and CT value of the bud (y = 1.9576 ln(x) −2.1655; R2 = 0.9691). Conclusion: CT technology can be used for visualization and quantitative analysis of the internal structure of the coconut, and some morphological changes and composition changes of the coconut during the germination process were observed during the three-month experiment. Therefore, CT is a potential tool for analyzing coconuts. [ABSTRACT FROM AUTHOR]