Search

In recent years, research based on anchor-based two-stage detectors has achieved great performance improvements in aerial object detection tasks. However, they still have two significant problems in the detection of tiny objects: 1) The preset fixed anchor is not conducive to assigning positive and negative samples in RPN when dealing with tiny objects, resulting in low-quality samples. 2) When the detector encounters tiny objects lacking structural details, it fails to accurately represent features, causing divergence in object features and hindering network learning. In this work, we propose the Anchor Adaptation and Feature Enhancement Strategies (AFS) to alleviate the above two problems. AFS contains two optimized modules: Anchor Adaption RPN Head (A2RH) and Feature Enhanced Attention Module (FEAM). Specifically, A2RH performs anchor adaptive learning by establishing a new anchor bias learning branch from the feature map, enabling higher-quality positive and negative sample assignments in RPN. FEAM introduces global features and mask attention based on FPN, and presents Gaussian mask supervision for attention to obtain stronger feature representation. Experiments show that our method improves the average precision by 1.8% on the baseline model, and achieves state-of-the-art results on AI-TOD dataset. Moreover, validation on AI-TOD-v2 and VisDrone2019 datasets also confirms the effectiveness of our method. The code will soon be available at https://github.com/gravity-lhg/AFS.

The technology of military equipment entity extraction, a crucial component in constructing military knowledge bases, holds significant research value and theoretical importance for guiding the development and improvement of equipment support forces. In the military domain, equipment entities exhibit a phenomenon of nesting, where one entity is contained within another, and abbreviations or codes are frequently used to represent these entities. To address this complexity, this paper proposes a method named CoTNER for extracting entities. Initially, a large-scale language model is used to perform data augmentation with chain-of-thought on the original dataset, providing additional semantic and contextual information. Subsequently, the augmented dataset is fine-tuned on a small-scale language model to adapt it to the task of military equipment entity extraction and to enhance its ability to learn complex rules specific to the domain of military equipment. Additionally, a high-quality data filtering strategy based on instruction-following difficulty scoring is proposed to address the catastrophic forgetting issue that may occur during the fine-tuning of large language models. The experimental results demonstrate that the proposed military equipment entity extraction method outperforms mainstream traditional deep learning methods, validating the effectiveness of CoTNER.

Recurrent neural network models have problems such as memory loss and gradient disappearance when dealing with long time series data. This paper proposes a long-term trajectory prediction model based on Transformer to process long-term sequence information. Firstly, the position encoding is used to preserve the relative positional relationship between trajectory points. Secondly, the multi-head attention mechanism is used to fully learn the feature information between different trajectories, and the trajectory data can be encoded at one time. Finally, the encoder and decoder mechanism is used to predict future trajectory data. Compared with the long-term trajectory prediction benchmark method TrajAirNet, the average displacement error, absolute displacement error of the proposed model on the long-term trajectory dataset are reduced by about 8.2% and 51.4%, respectively. The experimental results show that the proposed model has higher accuracy and robustness on long-term trajectory prediction dataset.

Abstract The data of coal mine safety field are massive, multi-source and heterogeneous. It is of practical importance to extract information from big data to achieve disaster precaution and emergency response. Existing approaches need to build more features and rely heavily on the linguistic knowledge of researchers, leading to inefficiency, poor portability, and slow update speed. This paper proposes a new relation extraction approach using recurrent neural networks with bidirectional minimal gated unit (MGU) model. This is achieved by adding a back-to-front MGU layer based on original MGU model. It does not require to construct complex text features and can capture the global context information by combining the forward and backward features. Evident from extensive experiments, the proposed approach outperforms the existing initiatives in terms of training time, accuracy, recall and F value.

Information used in existing ontology matching solutions are usually grouped into four categories: lexical information, structural information, semantic information, and external information, respectively. By summarizing and analyzing the approaches for utilizing the same kind of information, this paper finds that lexical information is mainly analyzed based on text and dictionary similarity. Similarly, structural information and semantic information are mainly analyzed based on graph structure and reasoner, respectively. The approaches for aggregating information analysis results are discussed. Challenges in the analysis of various types of information for existing ontology matching solutions are also described, and insights into directions for future research are provided.

The performance of software fault localization techniques is critical to software debugging and the reliability of software. Spectrum-based fault localization (SBFL) and mutation-based fault localization (MBFL) are the two most popular fault localization methods. However, the accuracies of the two methods are still limited. For example, only 10.63% of faults can be detected by inspecting the top 3 suspicious elements reported by Ochiai, which is a famous SBFL technique. Unfortunately, programmers only examine the first few suspicious elements before losing patience. Since the information used in SBFL and MBFL are quite different and complementary, this paper proposes a novel approach by combining spectrum and mutation to improve the fault localization accuracy. First, the faulty program is evaluated by using SBFL, and the potential faulty statements are ranked according to their suspiciousness. Then, mutants of the program are generated and executed by MBFL. Finally, the statements that are ranked in the top tied ${n}$ by SBFL are evaluated and reranked according to their mutation scores. Experiments are carried on the Defects4J benchmark and the results reveal that the accuracy of the proposed approach outperforms those of the SBFL and MBFL techniques. In terms of the faults located by inspecting the top 1 suspicious elements, the SMFL techniques detect at least 2.36 times more faults than two SBFL techniques (DStar and Ochiai) and detect at least 1.86 times more faults than two MBFL techniques (MUSE and Metallaxis).

Fuzz testing has been successful in finding defects of various software packages. These defects include file parsing, image processing, Internet browsers, and network protocols. However, the quality of the initial seed test cases greatly influences the coverage and defect detection capability of fuzz testing. To address this issue, we propose CSEFuzz, a fuzz testing approach based on symbolic execution for defect detection. First, CSEFuzz generates candidate test cases by symbolic execution and collects coverage information of the test cases. Then, CSEFuzz extracts the test-case templates of the test cases and selects a set of test-case templates according to specific coverage criteria. Finally, CSEFuzz selects test cases according to the selected test-case templates, and the selected test cases are used as initial seed test cases for fuzz testing. Experiments are conducted on 11 open-source programs. The results show that in comparison with afl-cmin, which is the test-case selection command of Kelinci, CSEFuzz with a path coverage criterion reduces the time costs of the initial seed test selection and verification by 94.26%. In addition, compared with afl-cmin, 32 more paths are covered and 16 more defects are detected by CSEFuzz.

Deep neural networks have been used for traffic classifications and promising results have been obtained. However, most of the previous work confined to one specific task of the classification, where restricts the classifier potential performance and application areas. The traffic flow can be labeled from a different perspective which might help to improve the accuracy of classifier by exploring more meaningful latent features. In addition, deep neural network (DNN)-based model is hard to adapt the changes in new classification demand, because of training such a new model costing not only many computing resources but also lots of labeled data. For this purpose, we proposed a multi-output DNN model simultaneously learning multi-task traffic classifications. In this model, the common knowledge of traffic is exploited by the synergy among the tasks and improves the performance of each task separately. Also, it is showed that this structure shares the potential of meeting new demands in the future and meanwhile being able to achieve the classification with advanced speed and fair accuracy. One-shot learning, which refers to the learning process with scarce data, is also explored and our approach shows notable performance.

CHn is the precursor unit for graphene synthesis. We have theoretically predicated a series of CHn structures with n = 1, 2, 4, 6, 8, 10, and 12 at elevated pressures (ambient pressure, 50, 100, 200, 300, 350, and 400 GPa) using evolutionary algorithms. The predicted CH and CH2 structures are graphane-type and polyethylene over the whole considered pressure range, respectively. The molecular crystalline methane is predicted for the stoichiometry of CH4. The combination of methane and H2 for CH6, CH8, CH10, and CH12 up to 300 GPa are obtained. At 400 GPa, the mixture of polymer and H2 for CH6, CH10, and CH12 comes into play. From the computed enthalpy, higher pressure and more hydrogen concentration contributed to the decomposition (to carbon and H2) of CHn systems. The total density of states for these CHn structures show that only the CH12 phase is metallic above 300 GPa. The rotational properties are traced in H2 and the CHn structures. The CH4 rotation is more sensitive to the pressure. The H2 units are nearly freely rotational. Other structures of CHn, including fcc-type and experimentally known structures, are not competitive with the structures predicted by evolutionary algorithms under high pressure region. Our results suggest that the CHn (n > 4) system is a potential candidate for hydrogen storage where H2 could be released by controlling the pressure.

Internet of Things is smartly changing various existing research areas into new themes, including smart health, smart home, smart industry, and smart transport. Relying on the basis of “smart transport,” Internet of Vehicles (IoV) is evolving as a new theme of research and development from vehicular ad hoc networks (VANETs). This paper presents a comprehensive framework of IoV with emphasis on layered architecture, protocol stack, network model, challenges, and future aspects. Specifically, following the background on the evolution of VANETs and motivation on IoV an overview of IoV is presented as the heterogeneous vehicular networks. The IoV includes five types of vehicular communications, namely, vehicle-to-vehicle, vehicle-to-roadside, vehicle-to-infrastructure of cellular networks, vehicle-to-personal devices, and vehicle-to-sensors. A five layered architecture of IoV is proposed considering functionalities and representations of each layer. A protocol stack for the layered architecture is structured considering management, operational, and security planes. A network model of IoV is proposed based on the three network elements, including cloud, connection, and client. The benefits of the design and development of IoV are highlighted by performing a qualitative comparison between IoV and VANETs. Finally, the challenges ahead for realizing IoV are discussed and future aspects of IoV are envisioned.

Iron sulfide nanoparticles play an essential role in the global biogeochemistry sulfur cycle and the origin of life. Controlling phase and morphology is essential but a great challenge for nanomaterials. Herein, we have investigated surface energy, morphology, thermodynamic stability, and phase evolution of FexSy nanoparticles through density functional theory calculations and Wulff construction. The sulfur-rich surfaces become more stable under higher sulfur chemical potential with low temperature and high partial H2S pressure, while the trends of sulfur-poor surfaces are reversed. Thermodynamic equilibrium phase diagrams imply that the intrinsic iron is sulfurated to mackinawite (FeS) and further to the polymorphs of FeS2 when the particle size is larger than 5 nm. When the size is decreased to 2 nm, the mackinawite (FeS) could be sulfurated to pyrrhotite (Fe7S8) and greigite (Fe3S4) and finally to the pyrite (FeS2) phase. [ABSTRACT FROM AUTHOR]