最新刊期
卷 52 ， 期 4 ， 2024
- 摘要：Approximately 71% of the Earth’s surface is encompassed by aqueous elements, such as rivers, lakes, and seas. Concurrently, terrestrial imaging contends with the influence of water in the forms of clouds, snow, rain, and fog. Notwithstanding, contemporary machine vision research and application systems predominantly concentrate on visual tasks within aerial and vacuum environments, leaving a dearth of systematic investigation into visual tasks within various aquatic contexts. Water-related vision, emblematic of water-based optical technology in the realm of vision, is committed to dissecting the scientific intricacies of light-water interactions and their inter-medium propagation. It also entails intelligent processing and analysis of visual image signals within aquatic settings. This discipline concurrently addresses engineering and technical intricacies intrinsic to the progression of advanced, intelligent water-related vision apparatus. Embarking from the fundamentally significant scientific query, “What is the reason for the ocean’s blue color?” this paper proffers an exhaustive survey encapsulating the repercussions of seawater’s light absorption, scattering, and attenuation mechanisms upon underwater visual tasks. Furthermore, the current methodologies for the processing and refinement of subaquatic images are systematically examined. Exploiting the optical attributes of water and factors contributing to image degradation, this manuscript underscores our team’s milestones in pioneering indispensable technologies for underwater imaging and image analysis. Substantial headway has been achieved in devising underwater observation and analytical apparatus, encompassing the full-ocean-depth ultra-high-definition camera “Haitong,” the full-ocean-depth 3D camera, and the full-ocean-depth high-definition video camera. These innovations have distinctly established a comprehensive and methodical proficiency in optical detection within submerged contexts, encompassing variables of color, intensity, polarization, and spectral analysis. This collective endeavor effectively bridges the gap in China’s full-ocean-depth optical detection technology, propelling the progress of exploration and technological innovation within the domain of water-related vision, which offers remarkable application value and societal advantages.关键词：water-related vision;water-related optics;multi-modal cognitive computing;machine vision;image and video signal processing;extraterrestrial oceans57|57|4更新时间：2025-12-08
- 摘要：The Internet, as a critical component of a nation's information infrastructure, has played a significant role in various domains. However, as its scale continues to expand and its applications deepen, we also face the potential catastrophic consequences of inconsistent network behaviors. To ensure the normal operation of the Internet and the consistency of network behaviors, there is an urgent need for deployable network verification technologies that align network operations with the intentions of network operators. Extensive research has been conducted on network verification technologies, assisting users in automating the detection of network errors and analyzing their root causes. However, to meet the increasing demands of the expanding Internet, scalability has become a crucial challenge in deploying network verification technologies. Specifically, how to quickly identify and diagnose errors in network policies, while satisfying time and space complexity constraints, has become a research hotspot in effectively applying network verification technologies in practice. To address this problem, this paper delves into and summarizes cutting-edge research on the temporal and spatial scalability of network verification. It begins by introducing the background knowledge related to network verification and then describes the current issues and challenges faced in network verification. Focusing on the core issue of scalability, the paper thoroughly analyzes existing work in achieving scalable verification from both the data plane and control plane perspectives. It provides a systematic analysis of the characteristics of these approaches, showcasing the distinctions and connections among related studies. According to the existing researches, we find that: (1) The scalability of data plane verification is primarily constrained by header space and forwarding matching rules, while the scalability of control plane verification is mainly limited by the complexity of multiple protocols and policies. (2) Although both data plane and control plane research employ similar scalable verification techniques, they address different but interconnected targets. For example, incremental computation in the data plane primarily focuses on updating packet equivalence classes, while incremental computation in the control plane primarily deals with network models affected by configuration changes. When applying network slicing techniques, both data plane and control plane independently validate the network by dividing it into multiple segments. (3) Compared to spatial scalability, current research places greater emphasis on temporal scalability, where reducing verification time overhead appears to be the primary pursuit of verification tools. (4) Previous research predominantly adopted a centralized verification approach, which involved collecting control plane or data plane information and then performing centralized analysis and verification. However, there has been a recent trend towards distributed verification, such as Coral and Tulkun in control plane verification. Lastly, based on the current research landscape, the paper concludes by summarizing and forecasting the research trends in scalable network verification technologies, offering valuable insights for researchers in this field. In conclusion, this paper presents a comprehensive review and outlook on the topic of scalability in network verification. It emphasizes the importance of aligning network behaviors with the intentions of network operators to ensure the reliable and consistent operation of the Internet. By addressing the challenges of scalability, researchers can advance the development of network verification technologies that can effectively verify large-scale networks within the constraints of time and space complexity. Ultimately, this contributes to enhancing the reliability and security of the Internet as a critical information infrastructure.关键词：network verification;scalability;network configuration;time-space optimization;data plane verification;control plane verification27|39|0更新时间：2025-12-08
- 摘要：Von Neumann computer architecture faces the bottleneck of “storage wall”, which hindering the performance improvement of AI (Artificial Intelligence) computing. Computing-In-Memory (CIM) breaks the limitation of “storage wall” and greatly improves the performance of AI computing. At present, CIM schemes have been implemented in a variety of storage media. According to the type of calculation signal, CIM scheme can be divided into digital CIM and analog CIM scheme. CIM has greatly improved the performance of AI computing, but the further development still faces major challenges. This article provides a detailed comparative analysis of CIM schemes in different signal domains, pointing out the main advantages and disadvantages of each scheme, and also pointing out the challenges faced by CIM. We believe that with the cross level collaborative research and development of process integration, devices, circuits, architecture, and software toolchains, CIM will provide more powerful and efficient computing power for AI computing at the edge and cloud ends.关键词：artificial intelligence;computing-in-memory;storage media;calculate signal type;evaluation index53|51|0更新时间：2025-12-08
- 摘要：When the working bandwidth exceeds 20 GHz (wavelength from less than 15 mm to infinitesimal), the electromagnetic (EM) environment of high-speed high-density electronic systems becomes very complex, and it is increasingly difficult to model, analyze and predict the EM response, and it makes the problem more acute when the bandwidth reaches more than 40 GHz. In order to be able to predict, evaluate and even control the process and effect of EM response at the design stage, accurate modeling methods and large-scale fast computing techniques are required, especially ultra-large bandwidth ultra-high-speed hybrid circuits and integrated circuits are involved. In this paper, a multi-scale processing technology based on the basic theory of electromagnetic computing in hybrid circuit environment is proposed, which solves the technical challenge of high-efficiency and high-precision field-circuit-hybrid modeling for electromagnetic environments of multi-scale complex circuits by employing the Huygens equivalence principle and the electromagnetic convergence and speed reduction mechanism of electromagnetic singularity. Using the Huygens equivalence principle and Kirchhoff integral equation, the Huygens port is defined on the regional boundary surface, and a general method of domain decomposition is proposed which can simultaneously cut entire region into vertical and horizontal subregions for arbitrary complex printed circuit board (PCB). The proposed realizes the hierarchical classification processing and modular packaging of any PCB structure, and improves the flexibility of high-speed high-density electronic system analysis. A technical approach based on eigen mode expansion method is proposed, and a fast parallel processing technology based on modal and domain decomposition and segmentation method is developed, which realizes both high precision and high calculation speed through the eigen-function representation for discontinuous field distribution, and reduces the calculation time and design time of complex electronic systems. Statistic data reveal frequency error of 3.7% and amplitude error of ±3 dB in verification range of 0~40 GHz. The hierarchical modeling and analysis method proposed in this paper can be applied to the design of high-end electronic communication system and improve the Chinese circuit design and control capabilities of the electromagnetic environment of broadband high-speed digital-analog system, and shorten the product development cycle.关键词：equivalent principle;eigenmode expansion;circuit modelling;domain decomposition;signal integrity;integrated circuit;package18|45|0更新时间：2025-12-08
- 摘要：Space communication is facing with harsh and dynamically changing electromagnetic environments. As the classical coding algorithm is designed based on the assumption of Gaussian channels, its direct application in space communication will result in low communication resource efficiency, poor flexibility, and transmission reliability deficiencies. This paper proposes a protograph-based generalized sparse random coding construction and implementation method for non-Gaussian channels in space communication. By introducing high-dimensional algebraic constraints on the protograph-based random coding architecture, the decoding performance at extremely low code rates is improved. This paper also proposes a multi-rate compatible coding structure for non-Gaussian channels and different application scenarios by dynamically adjusting the order and dimension of high-dimensional constraints. Finally, test results show that the proposed codes can achieve a large coding rate ranging from 1/40 to approximately 1, a variable information length ranging from 104 bits to 20000 bits, and a variable throughput ranging from 25 Kbps to 10 Gbps, which can achieve a frame error rate of 1.0×10‒4 at a low SNR of -37.1 dB (corresponding to
- 摘要：Tile-based methods that use the divide-and-conquer and on-demand transmission techniques are promising to handle 3D holographic video streaming. However, the current solutions either lack an adaptive tiling scheme or cannot apply to mobile real-time scenarios. In this paper, we propose VVSTiler (Volumetric Video Streaming Tiling selector), an adaptive tiling selector for holographic video communications, which can adaptively maximize perceived video quality under dynamic and limited computing and bandwidth resources. To be specific, we first conduct a preliminary study on the impacts of different tiling schemes and find that fine-grained tiles improve the rational utilization of dynamic network resources and coarse tiles ensure coding efficiency and robustness, which stimulates us to construct an adaptive tiling optimization based on the predicted viewport, available computing resources, and network bandwidth; and then devise a fast algorithm to enable online tiling decisions. Rich experiments on the 8iVFB (8i Voxelized Full Bodies) datasets are conducted to compare VVSTiler with state-of-the-art tiling-based baselines. The results exhibit that VVSTiler can achieve up to 60.4% video quality improvements and save on average 27% bandwidth per frame against the closest competitor, in cases of terrible and accurate viewport predictions, respectively.关键词：holographic video communications;volumetric video streaming;point cloud video;adaptive tiling;viewport prediction;perceived video quality30|29|0更新时间：2025-12-08
- 摘要：Program analysis plays a critical role in software development and maintenance. However, traditional logic-based program analysis methods exhibit significant limitations when dealing with modern, complex, large-scale, and dynamically rich software systems. The root cause of these limitations lies in the uncertainty present in software systems. To address this issue, researchers have proposed a series of new techniques for specific program analysis problems. These techniques combine probability information with traditional logic analysis to capture the uncertainty inherent in software systems. By summarizing and abstracting existing work in this area, this paper introduces the Bayesian program analysis framework. The core idea of this framework is to integrate program analysis with Bayesian statistical inference. It does so by modeling and updating probability distributions about the program to infer information about program behavior. Bayesian program analysis employs probabilistic logic programming to simultaneously handle both probability and logic information, providing a unified approach that encompasses various existing works. It can also be generalized to non-traditional static program analysis tasks, such as program fault localization and delta debugging. This paper provides a definition of the Bayesian program analysis framework, demonstrates its applications in program analysis and related fields, and outlines future directions for development.关键词：program analysis;logic programming;probabilistic logic programming;Bayesian network;bayesian inference36|40|0更新时间：2025-12-08
- 摘要：Microdroplet injection has a promising application in the field of printed electronics, which can complete the integrated molding of antenna dielectric substrates and conductive patterns in a single device. The surface topography quality of printed functional devices has a significant impact on their electrical properties. In this paper, for the problem of difficult to control the surface topographic quality in microdroplet jet 3D printing, a predictive control-based topographic compensation method for printed parts is proposed. Firstly, based on the layer-by-layer stacking behavior of the droplets, a topography prediction model of the printed part is established, which uses the matrix element update to describe the layer-by-layer evolution of the part topography, so as to accurately predict the existence of topographic defects in the multilayer printed part, such as the edge collapse and the large surface roughness. Then, a prediction controller is built based on this model to realize efficient compensation of topographic defects of the printed part by adjusting the print pattern of subsequent layers. Comparison experiments between open-loop printing and compensated printing are used to verify the effectiveness of the method. The experimental results show that the compensated printing method reduces the surface roughness of the printed part by 66.80% and the edge collapse by 43.22%, which effectively compensates for the surface morphology defects of the printed part. Finally, the microstrip patch antenna was fabricated using the microdroplet jet 3D printing process. The surface roughness of the dielectric substrate fabricated with the compensated printing method proposed in this paper is lower than that of the open-loop printed samples, which ensures the high-quality connection of the RF layer of the antenna. The return loss parameters were tested to be closer to the simulation results, thus demonstrating the significance of this study in printed electronics.关键词：three-dimensional printing;morphology evolution;microdroplet injection;print compensation;predictive control18|43|0更新时间：2025-12-08
- 摘要：With the development of ultra-low power (ULP) Internet of Things (IoT) systems, fast startup crystal oscillators with energy injection techniques have gained popularity due to their significant impact on the power consumption of IoT systems. Energy injection technology can significantly reduce the startup time and startup energy of the crystal oscillator but imposes strict restrictions on the accuracy of the injection source. To enlarge the injection frequency-mismatch tolerance and achieve high-efficiency injection, this work presents a delay locked-loop assisted phase-error correction technique. The injection frequency-mismatch tolerance is enlarged to 2%, and the non-injection duration of the start-up is only 4 cycles, achieving high injection efficiency. The proposed crystal oscillator is designed and fabricated in a 40 nm CMOS process. The prototype is tested with a 24 MHz crystal under 1.0 V supply. It achieves 7.2 μs startup time and consumes 5.1 nJ startup energy under 2%-injection-frequency-mismatch injection. Compared to the traditional injection with 2% injection frequency mismatch, this work achieves a startup time shorted by 99.66%.关键词：phase-error correction;crystal oscillator;fast startup;delay-locked loop;energy injection17|44|0更新时间：2025-12-08
- 摘要：Low-level complex wind field usually refers to the wind field below 600 m, with rapidly fluctuating speed and direction. The fine-resolution and large-scale detection under all-weather conditions is a long-term challenge in radar science, meteorology, etc., involving key technologies such as sensor design, three-dimensional wind field retrieval and information fusion. This paper reviews the technical connotation and research status of wind field detection from the aspects of light detection and ranging (LiDAR) detection, microwave radar detection, all-weather data fusion and hazard feature extraction, and finally analyzes its development trend. We believe this work provides reference for the development and application of all-weather precision detection technology in low-level complex wind field.关键词：low-level complex wind field;light detection and ranging (LiDAR);microwave radar;precision detection;all weather36|40|0更新时间：2025-12-08
- 摘要：The electric power infrastructure of China has developed into a highly informationized, automated, and intelligent cyber physical integration system. The interaction of cyber and physical not only significantly improves the efficiency and performance of power supply, but also introduces new network security threat. Cross-domain attacks that occurring in the cyber domain and acting on the physical domain can cause the systematic breakdown of power infrastructure and then lead to large-scale power outages. However, the current isolated cyber side or physical side defense system is difficult to effectively deal with these cross-domain attack threats.This paper introduces the current situation of information and physical cross domain attack threats faced by the power system, elaborates on the shortcomings of traditional defense methods in facing cross domain attacks, proposes a cross domain attack defense architecture based on information and physical collaboration, and designs defense methods from the perspectives of perception, identification, and blocking on the attack time scale. Through example design, it is proven that the proposed information and physical collaboration defense architecture can ensure the safe and stable operation of the power system.关键词：cyber physical power system;information-physical collaboration;cross-domain attack;coordinated defense;knowledge-data driven;power system safety44|44|0更新时间：2025-12-08
- 摘要：Element-level cooperation is the next stage of cooperative operation, and the dynamic reconstruction of kill chain is its typical feature. Aiming at the problem that platform-level cooperation is difficult to meet the demand for complex operational tasks, this paper proposes solution ideas and theoretical methods for heterogeneous platform element collaboration. Firstly analyzes the two main difficulties in realizing element-level cooperation, namely the complexity of element-level cooperation relationship and the uncertainty of cooperative ability; Secondly a based on feedback mechanism reconfigurable architecture, a dynamic reconstruction method for elements of the kill chain operator and a quantitative evaluation method for the effectiveness of collaborative order parameters were proposed through operational network modeling; Finally, The dynamic construction and reconstruction of the kill chain were achieved by simulation scenario, which can guide the algorithm design and engineering practice of element collaboration.关键词：element-level cooperation;cross-domain operations;dynamic refactoring;system modeling;kill-chain operator;order parameters22|39|0更新时间：2025-12-08
Special Issue: Recipients of CIE Science and Technology Awards
- 摘要：At present, programming abstractions of polymorphic network (PINet) are tightly coupled to the underlying hardware, and thus programs are monolithic and target-specific. As a result, it is inefficiency and error-prone to develop programs continuously, which compromises the availability and reliability of hardware devices. In this paper, we present the PINet’s programming environment (PPE) that aims to enable incremental development of protocols and functions. Based on the abstraction of one-big-switch, PPE proposes a target-independent model & language to support modularity and portability. It improves development efficiency by powerful forms of program composition. Correspondingly, the compiler framework is separated from front-end and back-end, so that it can automate the composition and analysis of the distributed programs. The packet processing logic is optimized to fit the resource constraints of hardware. The experimental results show that, PPE can reduce the lines of code by 20% without the affect to hardware performance, while introducing reasonable compilation delay and resource overhead.关键词：programming abstraction;polymorphic network;programmable data plane;modular;incremental programming;network modality13|45|1更新时间：2025-12-08
- 摘要：High-precision time-frequency comparison is a crucial demand in modern science and technology. Yet, the performance of current technologies is significantly hindered by the measurement capabilities of comparison systems and the error processing algorithms of link systems. Addressing these challenges, this study establishes a high-precision satellite-ground microwave time-frequency comparison system based on the China Space Station (CSS), introducing a novel satellite-ground time-frequency comparison technique that employs two-way measurements (tri-frequency mode) and includes advanced methods for detailed modeling of link system errors. Simulations across various satellite-ground time-frequency comparison scenarios reveal that, even with high-precision clocks (10-15 to 10-17), the long-term performance of the comparison, optimized through our algorithm, can achieve near-original satellite-ground clock performance. Preliminary comparison tests using the hydrogen maser clocks on the CSS and the ground demonstrate that, within a 300-second visibility arc, CSS-ground time-frequency comparison precision can reach 10.77 ps, with an ADEV (Allan Deviation) of 9.992 1×10-14@100 s, closely matching simulation results. This research provides new theoretical foundations and technical references for precision time-frequency synchronization technology.关键词：China Space Station;atomic clock;satellite-ground microwave link;time-frequency comparison;two-way measurement32|44|0更新时间：2025-12-08
- 摘要：The existing side-channel protection schemes for the uBlock algorithm suffer from high latency, making them unsuitable for low-latency and high-throughput scenarios. Additionally, these schemes lack provable security under the glitch-extended probing model. To address these issues, this paper presents a low-latency Threshold Implementation of the uBlock algorithm with provable security under the glitch-extended probing model. Furthermore, we introduce the Changing of the Guards technique to eliminate the need for additional random numbers during the execution of the protection scheme. To validate the security of our protection scheme, we employ the automated evaluation tool SILVER to assess the glitch-extended probing security of the S-box and utilize the leakage evaluation technology TVLA (Test Vector Leakage Assessment) to verify the security of the entire circuit. Finally, we evaluate the performance overhead of our protection scheme using the design compiler tool. The evaluation results demonstrate that our scheme achieves a significant reduction in latency, approximately 95% less compared to serialized implementations of uBlock protection schemes.关键词：cryptographic chip;uBlock algorithm;side-channel attack and protection;masking;threshold implementation;changing of the guards19|54|0更新时间：2025-12-08
- 摘要：In view of the problems of high demand of computation offloading in vehicular networks and the randomness of topology variation of vehicular networks, this paper proposes a vehicular multi-hop clustering scheme based on neighbor following and investigates its stability by multiple indicators. Existing research offers various vehicular clustering schemes, but they suffer from issues such as single indicator, being limited to a single hop or fixed number of hops, low clustering efficiency, and lacking a mechanism for cluster head replacement. As the neighbor following strategy has better stability and higher clustering efficiency, in the scheme designed in this paper, vehicles should follow the most stable vehicle based on the comprehensive indicators within neighboring nodes. The de-ringing and pruning algorithms are used to standardize and flatten the following structure of vehicles, forming a more stable vehicular cluster. The master-slave cluster header and the cluster maintenance mechanism are employed to enhance the robustness of vehicular cluster. Simulation results show that the proposed algorithm outperforms existing methods in terms of cluster stability and clustering efficiency.关键词：vehicular networks;neighbor following;multi-hop clustering;networking stability;state transition;clustering efficiency21|35|0更新时间：2025-12-08
- 摘要：In order to study the effects of interfering vehicles and crossing pedestrians on vehicle-to-vehicle visible light communication system, a vehicle-to-vehicle 2×2 MIMO visible light communication model with external interference is proposed in the paper. In this model, the automotive headlights based on the market-weighted model are considered as two transmitters and the photodetectors as two receivers. The effects of vehicles in the spaced lanes and pedestrians are considered. The results show that when the bit error rate is less than 10-4, the longitudinal communication distances of vehicles in the same lane, adjacent lane and spacing lane are 28 m, 5~18 m and 0 m, respectively. The two vehicles in the separate lane cannot communicate directly, but the light emitted by the headlights will enter the signal receiving vehicle and affect the normal communication between the two vehicles. Pedestrian crossing the road will make the bit error rate of normal communication vehicles fluctuate greatly and have a positive effect. The larger the lateral distance between normal communication vehicles, the higher the positive effect proportion. The model proposed in this paper takes into account the interference of pedestrians and vehicles, and provides theoretical basis for the application of visible light communication in complex traffic networks.关键词：external interference;vehicle to vehicle;visible light communication;bit error rate16|41|0更新时间：2025-12-08
- 摘要：In order to solve the problem of low recognition rate of in-vehicle voice interaction in complex noise environment and difficult deployment on devices with limited computing resources, this article proposes a lightweight and robust voice recognition method based on joint training framework in the noisy environment. The speech enhancement model introduces a multi-scale channel time-frequency attention module to extract multi-scale time-frequency features and key information in various dimensions. In the speech recognition model, multi-head element-wise linear attention is proposed, which significantly reduces the computational complexity required for the attention module. Experiments show that the joint training model shows good noise robustness on the self-made dataset.关键词：deep learning;speech enhancement;speech recognition;attention mechanism;joint training28|34|0更新时间：2025-12-08
- 摘要：Cryptographic ransomware extorts a ransom by encrypting user files. Existing early detection methods based on the first encryption-related application programming interface (API) cannot detect ransomware before it executes encryption behavior. Because the point at which different ransomware families begin executing their encryption behavior varies, existing early detection methods based on fixed time thresholds can only accurately detect a small fraction of ransomware before it executes encryption behavior. To further improve the timeliness of ransomware detection, this article proposes a concept that characterizes the time period from the start of software operation to the first call of encryption-related dynamic-link libraries (DLLs), namely the initial phase of operation (IPO). Based on the analysis of DLL and API call behavior in the early operational phase of several ransomwares, this article presents a method based on the API sequences generated by the software within the IPO as the detection object, namely the ransomware early detection method based on API latent semantics (REDMALS). REDMALS captures the API sequences within the IPO, uses the term frequency-inverse document frequency algorithm and the latent semantic analysis algorithm to generate feature vectors on the captured API sequences and to extract potential semantic structures, respectively, and then uses a machine learning algorithm to construct a detection model for ransomware detection. The experimental results show that REDMALS using the random forest algorithm achieves 97.7% and 96.0% accuracy on the constructed variant test set and unknown test set, respectively, and 83% and 76% of the ransomware samples in both test sets, respectively, can be detected before they perform any encryption behavior.关键词：ransomware;early detection;API;TF-IDF;latent semantic analysis;random forest38|39|0更新时间：2026-05-07
- 摘要：Accurate analysis of the left ventricular outline and ejection fraction through echocardiography holds significant diagnostic implications in cardiovascular diseases. However, current methodologies exhibit deficiencies such as a lack of correlation between left ventricular segmentation and ejection fraction prediction, susceptibility to outliers and abrupt variations in key points of left ventricular segmentation, substantial storage and computational overhead, and poor interpretability. In addressing these issues, this study proposes a lightweight graph convolutional method termed EchoGPK (Echo Guided by Priori Knowledge). Guided by prior knowledge encompassing cardiac structure, motion characteristics, and the similarity among adjacent myocardial regions, the approach incorporates a computationally efficient spiral aggregation function and a deeply compressed multi-head eccentric aggregation decoder, achieving the lightweighting of the graph convolutional structure. Leveraging the common experiences of clinical practitioners, the method introduces a multi task ejection fraction prediction network that moderately utilizes left ventricular contours, establishing a correlation between left ventricular segmentation and ejection fraction prediction to enhance interpretability. By employing the graph convolutional neural network transmission characteristics to constrain the behavior of neighboring points, the generation of boundary outliers and abrupt variations is reduced. Experimental results on the large-scale public dataset EchoNet-Dynamic demonstrate that EchoGPK achieves a Dice score of 92.13% for left ventricular segmentation and a mean absolute error (MAE) of 3.92% for ejection fraction prediction. Furthermore, the method exhibits higher accuracy, superior parameter count and computational efficiency compared to relevant approaches, affirming the effectiveness of prior knowledge in ultrasound medical image analysis.关键词：echocardiography;left ventricular segmentation;ejection fractions prediction;graph convolutional neural network14|35|0更新时间：2025-12-08
- 摘要：Diverse image captioning has become a research hotspot in the field of image description. Existing methods generally ignore the dependency relationship between global and sequential latent vectors, which seriously limits the performance improvement. To address this problem, this paper proposes a hybrid variational Transformer based diverse image captioning framework. Firstly, we construct a hybrid conditional variational autoencoder to effectively model the dependency between global and sequential latent vectors. Secondly, the evidence lower bound is derived by maximizing the conditional likelihood of the hybrid autoencoder, which serves as the objective function for diverse image captioning. Finally, we seamlessly combine the Transformer model with the hybrid conditional variational autoencoder, which can be jointly optimized to improve the generalization performance of diverse image captioning. The experimental results on MSCOCO dataset show that compared with the state-of-the-art methods, when randomly generating 20 and 100 captions, the diversity metric m-BLEU (Mutual overlap Bilingual Evaluation Under study) has improved by 4.2% and 4.7%, respectively, while the accuracy metric CIDEr (Consensus based Image Description Evaluation) has improved by 4.4% and 15.2%, respectively.关键词：image understanding;image captioning;variational autoencoding;latent embedding;multi-modal learning;generative model12|50|4更新时间：2025-12-08
- 摘要：In order to track a non-circular source target under the influence of both the state errors of moving observer arrays and frequency jitter deviations, this paper proposes a direct tracking algorithm based on the CNN (Convolutional Neural Network) + BiLSTM (Bi-directional Long Short-Term Memory) neural network. The proposed algorithm first exploits the correlation between the received array signals of multiple moving observers in various frequency bands and the non-circular property of the radiation source signals. Therefore, an extended multi-station observation vector is established in the presence of modeling errors. Then, the non-circular signal subspaces of the extended multi-station observation vectors within multiple observing timeslots are used to form the spatiotemporal feature input sequence. Subsequently, the direct tracking model for a non-circular source target based on CNN and BiLSTM neural network is designed. After training this neural network, the trajectory vector for a non-circular source target in several timeslots can be directly determined. Because the proposed algorithm estimates the target trajectory vector directly from the subspaces of the original array signals, it has higher estimation accuracy compared to the traditional two-step tracking methods which extract measurement parameters and then estimate the trajectory from them. As the information of modeling errors can be learned from training the neural network, this algorithm can achieve the calibration for modeling errors. The simulation results show that the proposed algorithm has higher trajectory estimation accuracy compared to the traditional two-step tracking algorithm and existing direct tracking algorithm. It significantly improves the robustness of multi-station collaborative tracking under the influence of modeling errors.关键词：direct tracking;noncircular signal;modeling error;convolutional neural network;bi-directional long short-term memory22|28|0更新时间：2025-12-08
- 摘要：After transforming an optimization problem into an objective function, the degree of matching between the objective function and the chosen heuristic optimization algorithm determines the efficiency of the following optimization. By simulating multi-ground states evolution in adiabatic quantum computation, a fitness landscape exploration algorithm is proposed to reflect the optimization characteristics of the objective function and guide the selection of optimization algorithms and their parameters. In quantum ground state evolution, the ground state wave function of a particle tends to converge towards regions with lower potential energy, and the extent of convergence is influenced by the quantum effect strength. Using these features, we encode the potential energy field by the objective function in a multi-ground states evolution with diminishing quantum effect, and consequently the fitness landscape of the objective function is reflected by the distributions of a set of converging ground state wave function in this adiabatic evolution. Based on the quantum path integral, the algorithm is implemented using a downscaling diffusion Monte Carlo (DMC). Experiments illustrated that the algorithm comprehensively and intuitively reflected numerous features of the fitness landscape, and the obtained information could directly guide optimization thereafter. Its computational mode resembles that of heuristic optimization, as it does not introduce other computations during optimization. These features introduce a novel perspective to the study of fitness landscape.关键词：fitness landscape;heuristic optimization;adiabatic quantum computation;ground state evolution;diffusion Monte Carlo;quantum annealing10|28|0更新时间：2025-12-08
- 摘要：Dependence clusters are the largest sets of interdependent program components, where any change in one point of the cluster can trigger a chain reaction in other components. In practical production environments, detecting dependence clusters is of great importance for software understanding, testing, and maintenance. Traditional dependence cluster detection methods rely on the system dependence graph (SDG) to calculate dependencies, However, constructing an SDG is a complex process that incurs significant time and space costs when analyzing large programs. In order to improve the efficiency of cluster detection, this paper proposes a light-weight and efficient dependence cluster detection method based on higher-order functions, which can avoid constructing SDG in the analysis. This method constructs a function summary in the form of a higher-order function for each procedure, where the data dependencies of formal parameters and global variables are used to initialize the dependence inside the procedure. The dependence cluster information between procedures can be obtained by instantiating the function summary and passing the dependence through summary parameters at the call site, which avoids the construction of an SDG. To further improve the analysis efficiency of the higher-order function-based detection method, we propose an optimization strategy based on adaptive computing, which significantly reduces the redundant calculations caused by the mutual recursive calls between functions. In the end, we select benchmark test sets with different scales and fields and conduct relevant experiments, which demonstrate that the proposed program dependence cluster detection method based on higher-order functions can improve analysis efficiency by 268.9% and reduce space usage by 35.7% compared with the detection method based on SDG. The optimization strategy based on adaptive computing can reduce redundant calculations by 56.7% and improve analysis efficiency by 23.9% compared with the method based on higher-order functions.关键词：dependence cluster;higher-order function;adaptive computing;system dependence graph;formal concept analysis12|27|0更新时间：2025-12-08
- 摘要：As one of the main ideas to improve the artificial bee colony (ABC) algorithm, the superior individuals are used to enhance the exploitative capability of the solution search equation. However, in the related works, the fitness information is often considered as the sole criterion for evaluating the individuals, which may easily cause some problems, e.g., the premature convergence. In this work, an improved ABC variant is proposed based on multiple information guidance, called ABC-MIG. In ABC-MIG, three different solution search equations are designed by using the fitness, position, and similarity information, respectively, and these new solution search equations are used in different ways for the employed bee phase and onlooker bee phase. Meanwhile, to save the search experience for the scout bee phase, a modified neighborhood search strategy is used to handle the abandoned food sources. To verify the effectiveness of ABC-MIG, extensive experiments are carried out on the CEC2013 test suite and one real-world optimization problem, and six derivative algorithms and five well-known improved ABC variants are included in the performance comparison. The results confirm that ABC-MIG has very competitive performance, in terms of the result accuracy and convergence speed.关键词：artificial bee colony algorithm;superior individuals;multiple information;solution search equation;neighborhood search22|54|0更新时间：2025-12-08
- 摘要：For the permissioned blockchain, the commonly-used full-replication storage mode impacts the storage scalability and security. To solve the problem, a blockchain storage sharding scheme was proposed, which offered good performance, high reliability, and low storage overhead. The network is divided into several storage units, each of which holds a certain number of blockchain copies through cooperation among nodes. The initial allocation distributes each new block to fixed-number of nodes. According to the changing access pattern, the adaptive replication increases replicas of popular blocks, and the storage optimization switches unpopular blocks into encoded mode. Experimental results show that, compared with the full-replication mode, the proposed scheme reduces the storage overhead of full nodes by on average 71%, with the performance kept at a proper level.关键词：permissioned blockchain;blockchain scalability;distributed storage;data dispersion;storage sharding17|32|0更新时间：2025-12-08
-
Implicit Discourse Relation Recognition Integrating Feature Coding and Phrase Interaction Perception
摘要：Implicit discourse relation recognition is a challenging task because of its difficulty and universality. From the perspective of argument coding and argument interaction, an implicit discourse relation recognition model integrating feature coding and phrase interaction perception is proposed. The model considers both the characteristics of argument itself and the interaction characteristics between arguments, and optimizes separately. The part of argument coding incorporates bidirectional long short-term memory (BiLSTM) and recurrent attention convolution neural network (RACNN), which can capture global and local features of arguments in a more comprehensive way; in the part of argument interaction, the semantic relationship between arguments is modeled from phrase level, and a mechanism of phrase-level interactive attention is constructed. Also, neural tensor network (NTN) is used to dig into the relational pattern, which can better reflect the potential deeper relational relationship between arguments. Experimental results on penn discourse treebank (PDTB) dataset show that the F1 values of this model are superior to other comparison models.关键词：implicit discourse relation recognition;bidirectional long short-term memory;recurrent attention convolution neural network;phrase-level interactive attention;neural tensor network15|60|0更新时间：2025-12-08 - 摘要：In this paper, a new type of gallium nitride (GaN) /Ti3C2Tx composite material was synthesized by solvothermal method and nitriding method, and the sensing characteristics of trimethylamine (TMA) gas sensor based on GaN/Ti3C2Tx composite were studied. Through a series of characterization methods, the morphology and elemental composition of the composites were analyzed, and the successful composite of GaN and Ti3C2Tx was confirmed. The results show that GaN/Ti3C2Tx sensor can efficiently detect 1~200 ppm TMA gas at room temperature, and the detection limit of TMA is reduced from 10 ppm to 1 ppm compared with pure GaN sensor. In addition, the gas sensing results of the composite sensor further confirm its good anti-interference characteristics and long-term stability. The gas-sensitive mechanism of GaN/Ti3C2Tx composite sensor and the reason of its improved gas-sensitive performance compared with pure GaN sensor are explained by using the surface depletion layer model. Finally, a contactless TMA gas detection device was developed in this paper, and the test results show that the device is expected to achieve efficient real-time detection of exhaled air in patients with liver and kidney disease, and has great application potential in early screening of liver and kidney diseases.关键词：GaN/Ti3C2Tx;room temperature;trimethylamine;gas sensor;contactless detection81|25|0更新时间：2025-12-08
- 摘要：The well logging curve records the amplitude range of geophysical properties changing with depth and is the bond between well log and seismic data. It is also significant for reservoir lithology analysis and subsequent oil and gas exploration projects. However, instrument failure and other reasons will cause well-logging curves to be missing in the actual logging process. Re-logging is not only expensive but also difficult to achieve. Aiming at the problem that logging data is often missing during geological exploration, this paper proposes a logging curve reconstruction method based on the LSTM(Long Short-Term Memory)-attention model. At the same time, EMD-VMD (Empirical Mode Decomposition-Variational Mode Decomposition) decomposition is performed on the original logging signal and then the correlation between the components and the original curve is calculated. Some excrescent components are deletedto promote efficient and high-precision manual completion. This proposed method is applied to reconstruct missing logs acoustic (AC) and density (DEN), and the prediction results are compared with those predicted by LSTM and BP (Back Propagation) neural network. The results show that the LSTM-attention model has a better prediction performance, and the correlations between predictive and the original curves can reach 86.8% (AC) and 74.8% (DEN), higher than the traditional LSTM and BP neural network. After removing redundant signal components, the correlation coefficients increased by 1.4% (AC) and 4.0% (DEN). At the same time, the logging curve predicted by the proposed method has the lowest prediction error. Therefore, the representation learning based on LSTM with an attention mechanism has better prediction accuracy for well-logging curve reconstruction.关键词：long short-term memory;attention;log curves reconstruction;VMD;EMD17|33|0更新时间：2025-12-08
PAPERS
- 摘要：Internet is a crucial information infrastructure that promotes economic development and technology innovation. However, TCP/IP (Transport Control Protocol/Internet Protocol) architecture, i.e., the core technology that supports and regulates the operation of Internet, has remained almost unchanged for decades. This paper introduces the necessity of innovating Internet architecture from four aspects: the drawbacks of technology itself, the development of social economy, the transformation cycle of technology, and the revolution law of science and technology. From the perspective of the systems thinking, this paper elaborates the underlying causes of the disadvantages in TCP/IP architecture. Moreover, this paper leverages the systems thinking and reveals the nature of network functionality and the nature of network interconnectivity, and then clarifies the four properties of delivering information (i.e., object property, identity property, location property, and means property). Then this paper briefly introduces how CoLoR (Coupling service Location and inter-domain Routing) leverages the two natures and four properties to innovate the Internet architecture. Finally, this paper takes CoLoR as the example, and presents the basic idea of deploying novel Internet architectures.关键词：Internet architecture;heterogeneous networks;internetworking;inter-domain communication37|41|1更新时间：2025-12-08
SURVEYS AND REVIEWS
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09064830号-19
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0