Large Language Models (LLMs) have demonstrated strong capabilities in web search and reasoning. However, their dependence on static training corpora makes them prone to factual errors and knowledge gaps. Retrieval-Augmented Generation (RAG) addresses this limitation by incorporating external knowledge sources, especially structured Knowledge Graphs (KGs), which provide explicit semantics and efficient retrieval. Existing KG-based RAG approaches, however, generally assume that anchor entities are accessible to initiate graph traversal, which limits their robustness in open-world settings where accurate linking between the user query and the KG entity is unreliable. To overcome this limitation, we propose AnchorRAG, a novel multi-agent collaboration framework for open-world RAG without the predefined anchor entities. Specifically, a predictor agent dynamically identifies candidate anchor entities by aligning user query terms with KG nodes and initializes independent retriever agents to conduct parallel multi-hop explorations from each candidate. Then a supervisor agent formulates the iterative retrieval strategy for these retriever agents and synthesizes the resulting knowledge paths to generate the final answer. This multi-agent collaboration framework improves retrieval robustness and mitigates the impact of ambiguous or erroneous anchors. Extensive experiments on four public benchmarks demonstrate that AnchorRAG significantly outperforms existing baselines and establishes new state-of-the-art results on the real-world reasoning tasks.

The real-time data acquisition system (RTDAQS) in the bathymetric light detection and ranging (LiDAR) instrument, named “GQ-Cormorant 19”, previously developed by our group cannot completely acquire the echo data from the water surface and water bottom owing to the inaccurate determination of the instant of echo signal acquisition, resulting in echo data loss. Therefore, this study developed an adaptive echo signal acquisition method based on a field-programmable gate array (FPGA) chip. The proposed method utilizes the flying height for roughness adjustment, and the peak of echo signal detection and the fine adjustment are combined to accurately determine the instant of echo signal acquisition. The improved RTDAQS onboard the GQ-Cormorant 19 was mounted on unmanned aerial vehicle (UAV) and unmanned surface vessel (USV) platforms and the performance was verified through an indoor corridor and several outdoor water fields. The experimental results demonstrated that the improved RTDAQS can improve the effective echo data rate by 11.5% when compared with the previously developed RTDAQS. Therefore, it can be concluded that the improved RTDAQS can implement the adaptive high-speed acquisition of echo data and obtain high-quality echo data acquisition on various platforms, such as UAV and USV.

Existing photomultiplier tube (PMT)-based gating control systems embedded in bathymetric light detection and ranging (LiDAR) devices cannot automatically adjust the PMT gain in real time, resulting in the saturation distortion of echo signals or failure to receive echo signals. Therefore, this study proposes an adaptive adjustment method for the laser energy and PMT gain based on the self-feedback of echo data from a high-speed sampling and storage system. Zynq-7000 All Programmable SoC (ZYNQ) was used as the master controller, and the echo data analysis module, laser energy adjustment module, and PMT gain adjustment were designed based on the field-programmable gate array (FPGA). The corresponding control software was developed based on the ARM end to call the custom IP cores. The proposed method was experimentally validated using indoor corridors and outdoor water areas, such as pools, reservoirs, Wujiu Beach, and the Beibu Gulf. The experimental results demonstrated that the proposed method could adaptively adjust laser energy and PMT gain, with which the voltage amplitude fluctuation range of the echo signal was effectively reduced from the original -1.7 V–0 V (acquisition range of high-speed sampling and storage system) to -1.04 V– -0.26 V. With the comparative verification with multibeam sounder, the mean and standard deviation of the difference between the Z coordinates are -0.21 and 0.15, respectively, which indicates that the LiDAR using the proposed method achieves a similar accuracy of bathymetry.

Information cascading, referred as the phenomenon of an individual following the behavior of the preceding individual after observing its actions, is prevalent in real social networks and triggers intense research interests for the purpose of monitoring and controlling network epidemics. One of the typical lines of information cascading study belongs to the Influence Maximization Problem, which aims to algorithmically find the optimal seeds that can spread the information to the maximum number of nodes. Regardless of the tremendous efforts made in various algorithm design of finding such optimal seeds, it has not yet been well understood how the absolute influence power of the “optimal” source set affects the ultimate cascading, i.e., under which conditions the seeds are able or unable to influence an substantial fraction of the entire network. Most existing works have investigated the conditions of network scale influence under linear threshold model, where the activation of a node requires a large number of infected neighbors. Instead, in this paper we focus on the case of single source cascading, which is only possible to occur under the independent cascading model. We launch information cascading analysis from two aspects, i.e., the influence scale and network-scale cascading probability. Firstly, percolation analysis of the cascading outcome shows that estimating influence scale is equivalent to solving fixed point equations. Then, we investigate the speed and stability of information cascading based on fixed point analysis, which shows that the information cascading process almost surely terminates within logarithmic time complexity. Furthermore, the results are generalized to the stochastic block model, where we find that network-scale cascading is determined by the spectral radius of the community matrix. The analysis presented in this paper could help us better understand the conditions for different information cascading outcomes.

Endobronchial intervention is increasingly used as a minimally invasive means for the treatment of pulmonary diseases. In order to reduce the difficulty of manipulation in complex airway networks, robust lumen detection is essential for intraoperative guidance. However, these methods are sensitive to visual artifacts which are inevitable during the surgery. In this work, a cross domain feature interaction (CDFI) network is proposed to extract the structural features of lumens, as well as to provide artifact cues to characterize the visual features. To effectively extract the structural and artifact features, the Quadruple Feature Constraints (QFC) module is designed to constrain the intrinsic connections of samples with various imaging-quality. Furthermore, we design a Guided Feature Fusion (GFF) module to supervise the model for adaptive feature fusion based on different types of artifacts. Results show that the features extracted by the proposed method can preserve the structural information of lumen in the presence of large visual variations, bringing much-improved lumen detection accuracy.

In order to explore the land deformation and its causes in the coastal areas of Guangxi Beibu Gulf, 36 Sentient-1A satellite images were used to obtain the surface subsidence rate and cumulative deformation of Guangxi Beibu Gulf from January 2018 to December 2020 based on PS-InSAR technology. Combining night-light remote sensing and rainfall data, we analyzed the causes of surface deformation. The results reveal: The deformation space within 4 km of the coastline of Beibu Bay and the urban areas of Fangchenggang, Qinzhou and Beihai were mainly in the port area of Fangchenggang, Qinnan District of Qinzhou and Haicheng District of Beihai, with the maximum accumulated settlement of -174.3 mm and the average deformation rate of -22.664 mm/a. The land subsidence rate in the Beibu Gulf area was affected by human activities and groundwater level, and the cumulative subsidence was positively correlated with the intensity of human activities. The rate of surface deformation decreases during the rainy season in July and August every year.

A traditional bathymetric LiDAR (light detection and ranging) has disadvantages such as large volume, heavy weight, necessity for airport and runway, and high cost for operation. For these reasons, this paper presents an off-axis four-reflection optical structure for single-band (532 nm) bathymetric LiDAR carried on UAV (Unmanned Aerial Vehicle). This optical system fully considers characteristics of the laser echo energy under different water conditions, which relate with the optical system parameters, such as peak power of laser emission, field of view (FOV), receiver aperture area, etc. The proposed optical system designs the objective lens, which are composed of one APD detector and two PMT detectors, the primary mirror, the second mirror, the plane mirror and the third mirror, two split field mirrors that separate the echo signals from shallow water, medium water and deep water, respectively. This proposed optical system was verified in laboratory tank, swimming pool, Lijiang River, lake, and the Qiaogang Sea Bay. It is found that the maximum water depth measured can reach 25.0 m with an error less than 0.1 m averagely. The dimension and weight of this LiDAR reach 90mm×160mm×90mm, and 10.25 kg, respectively, which is lightest and smallest bathymetric LiDAR worldwide.

At present, there are many deep learning models in the identification field, but they usually have problems with identification accuracy and slow speed. In this paper, we propose the S-MobileNetV2+SegNet model to identify sugarcane, and improve the SegNet model by replacing the encoder VGG16 model with the reduced structure Mobile-NetV2 model and introducing different ratios of dilated convolution to expand the local receptive field to eliminate the problem of insufficient information capture. Then reduce the network structure of the decoder and the number of convolution kernels to achieve a network with fewer parameters. To verify the accuracy and speed of the S-MobileNetV2+SegNet model in sugarcane identification, it is compared with SegNet, DeepLabV3+, and DeepLab-V3+Mobile-NetV2 models. The experimental results show that the S-MobileNetV2+SegNet model has better results and performance for sugarcane identification.

In the research and development of the coastal ocean, the data of coastal water depth is very important. At present, the development of multispectral satellite water depth measurement is very rapid. In the actual water depth inversion, single-band method, double band method (ratio logarithm method), and multi-band method which are belonging to lyzenga's method, and Strumpf's logarithm ratio method, are widely used. But few people make a comparative analysis of these methods and study their differences. This paper uses the landsat8 oil data and multibeam depth data of Weizhou Island in Guangxi, preprocesses the oil data such as land water separation and atmospheric correction, and then retrieve the water depth by single band method, dual-band method(ratio logarithm method), multi-band method and Strumpf's logarithm ratio method, and uses standard deviation and R2 to evaluate the results of retrieving the water depth.

In networks where massive sources make observations of same entities, we intend to seek thetruth– the most trustworthy value of each entity from conflicting information claimed by multiple sources. Various methods are proposed for accurately inferring both source reliability and truths, yet relying heavily on centralized settings that incur tremendous overhead to source side. In this paper, we offer adecentralizeddesign of truth discovery task that can fit favorably to the environments with limited resources. Considering that sources forming the connected network and making individual observations, we undertake the joint maximum likelihood estimation (MLE) of truth and source reliability. Our decentralization framework simply allows each source to maintain local information exchange at a time, and computes very basic functions of data observations. To this end, we facilitate the decentralization by simplifying the MLE problem into optimizing an objective function. Upon the proof of NP-hardness, two proposed decentralized algorithms (exact and approximation) are decentralized and randomized via a combination of algorithms from their centralized counterparts that ensure performance guarantee. The derived time complexity features explicit data/network dependent terms, which leads to further acceleration in truth finding. Remarkably, in two well connected networks like random geometric and preferential attachment graphs, the accelerated approximation method enjoys logarithmic time complexity while preserving comparable accuracy to the centralized counterparts. The effectiveness of the proposed decentralizations are further empirically confirmed.

Distributed computing is known for its high efficiency of processing large amounts of data in parallel, at the expense of communication load between different servers. Coding was introduced to minimize the communication load by exploiting the repetitive computing, thus drawing great attention within the academia. Most existing works assume that all servers are identical in computational capability, which is inconsistent with practical scenarios. In this paper, we investigate a distributed computing system that consists of two types of servers, i.e., fast servers and slow servers. Due to the heterogeneous computational capabilities within the system, the overall computation time will be delayed by the slow servers, which is called the straggling effect. To this end, we develop a novel framework of coding-based distributed computing to alleviate the straggling effect. Specifically, for a given number of fast servers and slow servers with their corresponding computational capabilities, we aim to minimize the overall computation time by assigning different amounts of workloads to different servers. Further, we derive the information-theoretic lower bound of the communication load of the system, which is shown to be within a constant multiplicative gap to the achievable communication load by our scheme.