最新刊期
卷 49 ， 期 2 ， 2021
- 摘要：Aiming at the problem of high computational overhead and redundancy in LSTM model, this paper proposed a design method of LSTM accelerator based on data sparsity. This scheme uses Delta network algorithm to mine the sparsity of input data, and filters the multiplication of redundant matrix vectors without irregular loading of data. The calculation mode of matrix-vector multiplication is modeled to find the most efficient parallel array computing architecture design scheme. The experimental results on MINIST dataset show that when the filter threshold of Delta network algorithm is less than 0.5, the detection accuracy of LSTM neural network algorithm remains unchanged, and the computational performance is improved by 21.53 times.关键词：long short-term memory;field programming gate anay;sparseness;matrix-vector multiplication16|23|1更新时间：2025-12-08
- 摘要：Software defect prediction helps improve software quality and allocate software test resources reasonably. Many defect prediction models based on software metrics have been proposed. However, the existing software metrics are mainly focused on structure information of source code, and the semantic information is lacking. Compilation optimization is the result of deep analysis of program semantics, and intuitively we believe that it should reflect the semantic information of the program in some ways to help defect prediction. Based on the optimization options widely used in the current compiler, this paper extracts 9 compilation optimization metrics, and proposes five types of metrics models that designed by different metrics sets. The relationship between compilation optimization metrics and software defect predictions was evaluated by 13 commonly used classifiers in weka,and also compared with DP-CNN. Experimental results show:Compilation optimization metrics have a significant impact on the recall rate of software defect prediction; Static code metrics combined with compilation optimization metrics can improve the performance of software defect prediction in all classifiers, which can improve the performance of prediction by about 5%; Code size based optimization metrics and performance based optimization metrics have their characteristics, combined both of them can get better performance in software defect prediction.关键词：compilation optimization;software metrics;software defect prediction12|18|2更新时间：2025-12-08
- 摘要：A multi-objective optimization algorithm based on estimation of distribution algorithm, namely EDA (Estimation of Distribution Algorithm) is presented to minimize the maximum completion time and the total energy consumption for the green no-wait job shop scheduling problem (GNWJSSP). Double populations are used to increase the diversity of the population, and a heuristic operation named NEH (Nawaz-Enscore-Ham) is added to improve the quality of the initial solution. An adaptive learning rate is designed to control the guidance of the information of excellent solutions during the entire algorithm search process. Combined with the structural properties and decoding method of the no-wait job shop scheduling problem, a fast evaluation method based on the front omission strategy is proposed, and two local search strategies are constructed by utilizing the Insert operation, thereby enhancing the search efficiency and depth of the algorithm. The effectiveness of the proposed algorithm for solving GNWJSSP is verified by simulation experiments and algorithm comparisons.关键词：no-wait job shop scheduling problem;estimation of distribution algorithm;energy consumption;green scheduling5|22|4更新时间：2025-12-08
- 摘要：There are various deep learning methods already implemented in time-series forecasting problems, and some of them show better performance and adaptability than the methods based on statistics. However, the spatial dependence implicit in multiple series usually is not considered within the existing time-series forecasting methods, which also results in unsatisfactory performance in long-term forecasting. Thus, we propose a model series-decomposition-block recurrent neural network (SDBRNN), which fuses time-series decomposition strategy and the spatio-temporal convolutional layer. This model relies on an improved STL decomposition strategy to get optimal series-components and to combine them into spatio-temporal blocks. Then an long-short term memory (LSTM) variant network called Block-LSTM is used to extract spatio-temporal features from blocks and achieve forecasting. Experiments on real-world datasets proved that the model has excellent capabilities of feature extraction and long-term forecasting compared with other methods such as temporal convolutional network and recurrent neural network.关键词：time-series forecasting;series decomposition;spatio-temporal convolutional9|12|6更新时间：2025-12-08
- 摘要：The existing algorithms based on parity-check relationship in the recognition of interleaver for Turbo codes have weak adaptability at low signal-to-noise ratio (SNR) and are prone to "error propagation" in the recognition process, In order to further improve the fault tolerance and real time of interleaving recognition, we propose two error correction algorithms based on the maximum sequence correlation recognition algorithm, including iterative decoding and small traversal, which make full use of the received data and enhance the error resilience. Simulation results show that the performance of the improved algorithm is obviously better than the original algorithm by adding error correction algorithm, and it only needs 1/3 of the data to complete the identification of all the interleaved parameters under the same conditions and has more than 2dB gain to achieve the identification of all the interleaved parameters under the condition of the same amount of data. In addition, the method based on decoding error correction is universal, which is also applied to other recognition algorithms of parity-check relationship.关键词：Turbo-code;random interweaving;maximum sequence correlation;decoding;error correction17|5|3更新时间：2025-12-08
- 摘要：Domain-specific natural language understanding technology is one of the core technology of vertical search engines, domain-specific question answering system and other applications. This research focus on a novel constrained semantic grammar and its automatic learning methods based on an existing domain-specific question answering system. An error-driven learning method of semantic grammar is proposed. The method first partially parses the ungrammatical sentence based on the core semantic grammar, then it attempts to build a complete parse tree, including predicting the top-level node of the partial parsing tree, generating and verifying hypotheses of new grammar rules. Learnability metrics is used to filter sentences in the training corpus to improve the overall quality and efficiency of grammar extending algorithm. The proposed algorithm is applied to two domains of different scales. In the interactive learning paradigm, learning efficiency are compared in different domains. In the batch learning paradigm, the paper compares the accuracy, MRR and recognition rate of the extended grammar and core grammar on twodatasets. The test results show that the proposed method is effective.关键词：semantic grammar;grammar extending;natural language understanding;domain specific;ontology9|4|0更新时间：2025-12-08
- 摘要：Password-based authentication key exchange protocol is highly practical in modern communication networks. The rapid development of quantum technology has made the security of traditional public key cryptosystem face a severe situation. The construction of cryptosystem based on lattice theory has become a hot topic in the research of post-quantum cryptography. The scheme is based on the lattice theory error learning (RLWE) problem, and uses the Peikert error reconciliation mechanism to construct a password authentication key exchange (PAKE) protocol in C/S mode. Reasonable parameters are set to ensure that both parties get the same session key with significant probability, and by using Java to implement PAKE protocol on the Eclipse platform. The security of the protocol is proved under the standard BPR model and can withstand quantum attacks. Compared with the existing PAKE protocol based on lattice theory, the proposed protocol has lower communication and better security. It is a simple and efficient post-quantum password authentication key exchange protocol.关键词：authenticated key exchange protocol;ring learning with errors;lattice;post quantum5|16|10更新时间：2025-12-08
- 摘要：To better reveal the latent semantic correlation between image and text, this paper proposes a cross media retrieval method by fusing two level similarity, which constructs two subnets to deal with global features and local features respectively so as to obtain better semantic matching between image and text. The image representation is divided into whole image and some image regions, and the text representation is also divided into whole sentence and some words. A two level alignment method is designed to match the global and local representation of image and text, and the two similarities are fused to learn the complete cross-media representation. The experimental results on MSCOCO and Flickr30K datasets show that the proposed method can make the semantic matching of image and text more accurate, and is superior to many state-of-the-art cross-media retrieval methods.关键词：convolutional neural network;self-attention network;two level similarity;cross-media retrieval4|6|10更新时间：2025-12-08
- 摘要：In this paper the design and implementation of a new parallel adder, SJ-MSD adder, in a ternary optical computer (i.e.TOC) are proposed. It is introduced that a sufficient conditions (Shen's sufficient theorem) which judging that a group of ternary logical transformations is able to build a MSD (Modified Signed Digit) addition. Five logical transformations S1,S2,J1,J2 and J3 (i.e.SJ transformation) and a parallel carry-free MSD addition rule (i.e.SJ rule) are proposed to construct SJ-MSD adder. Meanwhile according to Shen's sufficient theorem, the reliability of the SJ transformation and SJ rule used to construct MSD parallel adder is proved. Subsequently, in the ternary optical computer prototype system SD16, the design and implement of SJ-MSD adder are discussed in detail. At the same time the implementation methods of pipeline addition and sharing one SJ-MSD adder for multiple data are expatiated. Some experiments of SJ-MSD adder is introduced. Compared with the previous TW-MSD adder, the SJ-MSD adder occupies processor bits less about 25%. So the optical processor’ efficiency is effectively improved.关键词：TOC(ternary optical computer);reconfiguration;pixel;processor bit;shareable pipelined adder4|13|2更新时间：2025-12-08
- 摘要：In recent years, automated sandboxes have been widely deployed for malware analysis and detection. However, with the rapid increase column of malware and the enhancement of anti-analysis capabilities, how to effectively handle these massive malware analysis tasks and improve the efficiency of sandbox system is an important research topic. Based on the characteristics of different learning methods and malware dynamic and static features, this paper proposes a malware detection method based on a hybrid learning model. We extract static fuzzy-hash features and dynamic behavior features of malware, then unsupervised clustering learning is combined with supervised classification learning. Experiments show that using only 0.02% of the analysis time improves the detection speed of the entire system by 25.6% without affecting the detection accuracy.关键词：malware;fuzzy hash;hybrid learning24|19|1更新时间：2025-12-08
- 摘要：Aiming at error and robustness of localization algorithms in underwater acoustic sensor networks, we proposed two monte carlo mobile localization algorithms: Circular ring monte carlo localization (CRMCL) and particle swarm optimization for circular ring monte carlo localization (PRMCL). CRMCL used one-hop anchor nodes to construct circular sampling area and ring filter. By defining sampling points density, we obtained reasonable sample number. The relationship between the ring parameter and the filtration area was demonstrated. Then the reasonable ring parameter was obtained through the simulation experiments, and an efficient filter was constructed to reduce the localization error. PRMCL used particle swarm optimization (PSO) algorithm to optimize the samples filtered by CRMCL, which reduced the number of invalid samples and improved the robustness of localization. The simulation results show that CRMCL and PRMCL have lower localization error and better robustness than monte carlo localization and other improved algorithms without additional hardware.关键词：underwater acoustic sensor networks;mobile localization;monte carlo localization (MCL);particle swarm optimization (PSO);ring filter11|17|3更新时间：2025-12-08
- 摘要：In massive machine type communication (MTC), when a large number of devices flood into the LTE-A (long term evolution-advanced) network in a short period of time,and simultaneously initiated random access attempt, which will cause serious network congestion. Appropriate measures need to be taken to control the congestion.Therefore, this paper proposes a congestion control scheme based on a sliding window. The size of the window controls the number of devices that simultaneously initiate attempts to reduce the collision probability between devices, which can effectively improve the success rate of system access and alleviate network congestion. In different load scenarios, the proposed scheme is compared with the receiving performance of the static ACB (access class barring) strategy and the dynamic ACB strategy. The simulation results show that the scheme proposed here can achieve access success rate close to 100%, while reduces the minimum average reception delay and overall system service time.关键词：massive machine type communication;LTE-A random access;congestion control;sliding windows;access success probability;average access delay15|14|0更新时间：2025-12-08
- 摘要：Aiming at the problem that mobile robots are prone to localization errors during global relocation, this paper proposes a particle filter relocation method based on the constructed semantic map and loading semantic likelihood estimation to solve the problem. Using the lidar to establish the environmental grid map, meanwhile combing with the three-dimensional depth camera’s object recognition and positioning information, the environmental semantic map is obtained by giving semantic information. During the relocation process, the particle filter method is used to simultaneously match the grid map structure and the semantic information of the environment to calculate the actual position of the robot on the map, and accurately to realize the position relocation. Experiments results show that this method can overcome the shortcomings of the original particle filtering method that only uses environmental structure information for matching, also solve the problem of robot global relocation error-prone and enhance the robustness of relocation, and enhances the convergence speed of relocation.关键词：global location;semantic map;particle filter algorithm;semantic likelihoo14|5|9更新时间：2025-12-08
- 摘要：To better learn node dependence and make use of structural information, this paper proposes a new method that takes the tree of complete dependence as the direct input. The method uses the graph convolutional network and combines two parallel attention modules to learn how to select the useful information. The method represents the samples as nodes on the graph. One module is used to compute the influence between positions of node features, which allows the feature vector to contain a wider range of semantic information. The other one is used to compute the relational features of node dependence, so as to enhance the global dependence between nodes. The two modules promote each other in parallel to obtain complete feature representation. The experimental results on the TACRED and SemEval datasets show that the method can obtain more useful information for relation extraction, thus achieve better performances on various evaluation metrics.关键词：relation extraction;graph convolutional network;attention mechanism;multi-hop relational reasoning82|22|7更新时间：2025-12-08
- 摘要：The flexible job shop scheduling problem is investigated, where the transportation resource and operation related setup time constraints are considered simultaneously. The objective is to minimize the maximum completion time and the energy consumption. To solve the problem, we propose an improved artificial bee colony algorithm, where each solution is represented by a two-dimensional vector, the scheduling vector is to record the operation processing sequence, and the machine assignment vector is to assign the candidate machine for each operation. In the decoding mechanism, the transportation and setup time constraints are investigated. For the local search approaches, we develop five types of neighborhood structures for the scheduling part, and a well-designed machine assignment neighborhood structure for the machine assignment vector. To enhance the global searching abilities, the simulated annealing acceptance method is embedded. Finally, the experiment comparisons verify the performance of the proposed algorithm.关键词：artificial bee colony;flexible job shop;energy consumption;crane transportation;setup time24|20|12更新时间：2025-12-08
- 摘要：The purpose of supervised learning for spiking neurons is to emit spikes at desired times. By analyzing the relationship between desired output spike train, actual output spike train, and input spike trains, we argue that the spike selection and calculation of existing algorithms are much complex. By removing the repetitive spike computations which have a negative effect on the overall learning performance, we construct a unit of pair-spike and propose a direct computation method of supervised learning algorithms for spiking neurons. Based on input spikes and the relationship between desired output spike train and actual output spike train, the proposed method utilities every useful input spike selected only once to directly adjust the synaptic weights, which truly reduces the computational cost. Experimental results show that the direct computation method, as a general optimization strategy of supervised learning for spiking neurons, can effectively improve the learning performance of existing algorithms.关键词：supervised learning;spiking neuron;spiking neural networks;spike train;direct computation147|18|2更新时间：2025-12-08
- 摘要：In order to attract users' attention and improve the efficiency of interaction, adjusting the position and increasing the area of interface elements is very common in adaptive interface. In this paper, we studied the cognitive characteristics of people in the process of the change of regional location and the increase of regional area of the adaptive interface, which provides guidelines for the design of the adaptive interface. Ergonomics experiments were designed to make the interface adaptive items randomly switch between the middle position and other locations, as well as the original area interface and the enlarged area interface. The subjects were asked to complete the visual search task, and the reaction time and accuracy of the subjects were recorded. The results show that the direction (up, down, left and right), and the distance have significant effects on the reaction time. The eye movement heatmap shows that subjects are more sensitive to change in the direction of the up and down. Under the experimental conditions, the adaptive change distance of the interface items in the up and down directions can not exceed 12.5% of the screen size; the change distance cannot exceed 25% of the screen size in the direction of left and right. When the increase is greater than 125% of the original area, the dynamic switching factor has a significant effect on the reaction time, so the proportion of area increase can not exceed 100%.关键词：ergonomics;adaptive interface;eye-tracking;cognitive characteristic10|10|1更新时间：2025-12-08
- 摘要：To improve the charging energy efficiency in wireless rechargeable sensor networks (WRSNs), this paper proposes an on-demand multi-node passer-by charging (OMPC) scheduling scheme. Mobile chargers are considered to be allowed to simultaneously charge unrequested nodes within the range of coverage while charging the request nodes. To further reduce the number of groups, a more effective heuristic algorithm is proposed compared with the existing algorithms, and the correctness of the selection of the stopping points is proved. In this paper, the effect of low-efficiency grouping on charging performance is analyzed, and an effective strategy for splitting low-efficiency grouping is proposed to optimize the travel path locally. Simulation results show that in offline and on-demand scenarios, the proposed algorithm and scheme can reduce energy consumption and improve charging efficiency.关键词：wireless rechargeable sensor networks;mobile charger;multi-node charging model;on-demand charging68|22|4更新时间：2025-12-08
- 摘要：The dynamic change of soil seepage field and its deep displacement are the main inducements of landslide geological disasters. To solve the problem that the existing deep displacement monitoring technology can not describe and explain the deformation process of landslide in space, an imaging prediction technology of landslide internal structure based on simplified interactive multiple model (SIMM) and unscented kalman filter (UKF) is proposed. Electrical prospecting has the characteristics ofcollecting underground media information quickly and accurately, Furthermore, it can realize multi-dimensional resistivity profile imaging. Based on the above and geological conditions, resistivity models collected by different exploration devices are taken as input. The threshold detection is used to judge whether the change speed of the resistivity value around the slip surface exceeds the warning value, and the data of each model are processed by unscented Kalman filter and output interactively. Field experiments show that the algorithm solves the problem of sensitivity matching between resistivity acquisition devices, realizes fast and multi-dimensional imaging of the internal structure of landslide geological body, and improves the effectiveness of interpretation of landslide internal deformation process.关键词：landslide geological hazard monitoring;resistivity exploration;interactive multi-model;unscented Kalman filter18|5|0更新时间：2025-12-08
- 摘要：Fabricated by degradable materials, transient electronics can dissolve in natural environments after completing the default tasks. Transient electronics can compare favorably with the traditional devices through advanced fabrication technology in terms of performance, exhibiting promising applications in fields of environmental protection, medical equipment and information security. This paper comprehensively analyzed its concepts, composition, fabrication technology, degradable process as well as transient devices, and the military application of transient electronic technology was expounded from four aspects: Safe storage of combat information, reconnaissance and surveillance of battlefield conditions, monitoring, auxiliary diagnosis and treatment of soldiers’ injuries, accurate delivery of urgently needed supplies. The existing problem in the development of transient electronic technology was pointed out, and its future development trend was prospected.关键词：transient electronics;trigger;biodegradable;military applications59|4|1更新时间：2025-12-08
- 摘要：With the development of deep learning, it has been competent to various types of tasks, such as identification, classification, and generation. However, the structures or parameters of artificial neural networks cannot be only a little changed for different task, experts need to adjust the structures or parameters of the neural network. In such situations, the method of automatically adjusting the structures or parameters of the artificial neural network has become a research hotspot, among these methods, neuroevolution inspired by Darwin's natural evolution theory has become the main optimization method for that. Deep learning models optimized by neuroevolution based on population, evolving through mutation, crossover and other operations, can automatically and gradually construct the neural network and then choose the most optimal deep learning model. This paper summarizes the neuroevolution and the evolutionary computation. It elaborates various deep learning models based on neuroevolution, and analyzes the performance of these models. It concludes prospects of the deep learning model based on neuroevolution and discusses the next research directions.关键词：neuroevolution;deep learning;evolutionary computation;convolutional neural networks;generative adversarial networks;autoencoder;long short-term memory networks;deep reinforcement learning120|12|7更新时间：2025-12-08
- 摘要：Hyperspectral microwave radiometer is based on the technological analysis of spectrum. It is able to increase vertical resolution of measurement of atmosphere by increasing frequency measurement channels. The back-end system of hyperspectral microwave radiometer sampled analogical signal with 4GHz frequency band at 10Gsps by using high-performance analog-to-digital converter (ADC). The output digital signal from ADC was dealt with on field programming gate array (FPGA) by implementing the algorithm of power spectrum. The detailed subsystem design, the ways to choose components meeting the design requirements and the testing methods were presented. In addition, some basic test results and analysis results were presented.关键词：technology of spectrum analysis;hyperspectral microwave radiometer;power spectrum;high-performance analog-to digital converter (ADC);field programming gate array (FPGA)14|16|1更新时间：2025-12-08
- 摘要：In order to solve the problem that differential local mean decomposition (DLMD) can’t adaptively determine the differential degree, a rolling bearing fault feature extraction method based on improved DLMD and Teager-Kaiser energy operator (TKEO) demodulation is proposed. Firstly, the index of DLMD differential degree based on the sum of midpoint local mean distance and absolute skewness is constructed, and the signal is decomposed into several product function (PF) components; Secondly, the sensitive factors are calculated, and the effective PF components were screened and reconstructed; Finally, the TKEO spectrum is calculated to extract the fault features of the rolling bearing. The experimental results show that the proposed method can adaptively judge the differential degree of DLMD and effectively extract the fault features of rolling bearing.关键词：differential local mean decomposition;rolling bearing;sensitive parameter;Teager-Kaiser energy operator9|6|2更新时间：2025-12-08
- 摘要：The feature size of CMOS technology is continuously shrinking, and the single event triple-node-upset induced by the charge sharing effect has become a research hotspot. This paper proposed a single event triple-node-upset self-recovery radiation-hardened latch: Hydra-DICE. The latch is arranged in an array structure based on 24 homogeneous cross-coupled elements (CCE). When a single event upset occurs at any three internal nodes concurrently, Hydra-DICE can realize the function of self-recovery to the correct logical value. Compared with the TNURL latch which has the equivalent triple-node-upset self-recovery capability, the Hydra-DICE latch has a 50% reduction in area overhead, a 48.28% reduction in delay, a 25.00% reduction in power consumption, and a 61.21% reduction in power consumption delay product. The simulation results show that the hardened latch has made a good compromise in fault tolerance performance, area overhead,delay and power.关键词：latch;single event upset;dual interlocked storage cell;radiation-hardened;self-recovery12|2|1更新时间：2025-12-08
- 摘要：Redundant expressions, misuse of words, and missing content and other text errors can seriously affect the interpretation of text semantics. There exist two major problems with current text error correction models: The Encoder-Decoder based text error correction models have slow decoding speed; text error detection task and text correction task are handled as two separate tasks. Hence, a text error correction model based on a hierarchical editing framework is proposed in this paper. Firstly, a variety of text semantic representations are obtained through modelling pre-trained model. Secondly, text errors are located by using these text semantic representations. Finally, on the basis of hierarchical editing framework, precise editing operations are worked out to edit the errors. Experiments on the published text error correction dataset show that the proposed model has faster decoding speed and higher recall rate than comparison models.关键词：text error correction;pre-trained model;hierarchical editing framework;deep learning7|22|5更新时间：2025-12-08
- 摘要：Aiming at the problem that the number of commercial Wi-Fi access points (AP) antennas restricts the high-precision positioning based on angle of arrival (AoA), this paper proposes an indoor real-time positioning algorithm based on Wi-Fi signal. An AoA estimation algorithm using Wi-Fi is proposed, which can quickly estimate the AoA of the line of sight (LoS) path with few antennas and snapshots to ensure real-time positioning. The IEEE 802.11 Saleh-Valenzuela (S-V) channel model analyses the influence of multipath signals on the energy spectrum peaks of direct signals. In order to improve the positioning accuracy, a multi-AP joint positioning algorithm based on antenna selection is proposed. A real-time positioning demonstration system is built to verify the effectiveness of the system. The experiment results show that the proposed algorithm can achieve 67% 1.2m positioning accuracy and the positioning delay is less than 0.5 seconds.关键词：indoor localization;S-V channel model;AoA estimation;Wi-Fi16|8|4更新时间：2025-12-08
- 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