最新刊期
卷 30 ， 期 1 ， 2002
- 摘要：We systematically discuss a high-dimensional space geometry method used for analyzing artificial neural networks,do some research work about the high-dimensional space geometry and prove the necessary theorems.It is a novel tool to analyze artificial neural networks.关键词：neural networks;high-dimensional space;learning algorithm;high-dimensional geometry2|2866|64更新时间：2025-07-16
- 摘要：In this paper,we employ wavelet networks for demodulation of DS-CDMA(direct sequence-code disivion multiple access) signals in a multi-access environment.In practice,conventional detector (single-user matched filter) suffers severe performance degradation as the relative powers of interfering signals become large.On the basis of MMSE(Minimum Mean-Square Error) linear detector we use adaptive wavelet network to achieve minimum mean square error,and effectively eliminate multi-access interference.Simulations show the detector based on wavelet networks can converge rapidly and is near-far resistant.关键词：DS-CDMA system;Multiple-access interference;multi-user detection;wavelet transform;wavelet network2|612|0更新时间：2025-07-16
- 摘要：An analysis of CDMA downlink capacity is presented in this paper.The analysis is derived from the basic CDMA theory,and is then fitted into the scenario in question with a help of a number of reference simulations.Some results of analysis and simulation are presented.Comparison with the system simulations shows that the analysis result follows well the simulation results.关键词：CDMA;capacity;downlink;power limited2|1137|2更新时间：2025-07-16
- 摘要：The extended characteristic equation methods are widely used by most harmonic retrieval approaches.While the false peaks caused by these methods disturb seriously to distinguish the false peaks from the harmonic signal peaks.This paper has successfully solved the problem by studying the singular value decomposition (SVD) natures of the cross-correlation matrix.Simulation results show that this new cross-spectral SVD-LS approach can clear spectral false peaks.关键词：harmonic retrieval;false peaks;extended equation2|1217|8更新时间：2025-07-16
- 摘要：A new distributed secure electronic auction protocol based on group signatures and Shamir's threshold secret sharing scheme is proposed.This protocol is designed for a loosely coupled broadcast/subscribe architecture.The protocol not only guarantees the non-repudiation and anonymity of the bidder,but also ensures the non-repudiation of the bid received.Compared with previous work,our protocol can provide more security,and is suitable for large scale distributed auction.关键词：distributed systems;security;electronic auction;broadcast/subscribe architecture;group signatures2|921|5更新时间：2025-07-16
- 摘要：The HMM based multiple frequency line tracking algorithms of work well under very low SNR,except for a high complexity of computation and for generally unnegligible quantization error.In this paper a new choice of measurement vector and the corresponding computation of measurement probability are presented to reduce the quantization error.And to reduce the computation burden,the idea of the EM algorithm is applied to the multiple frequency line case and the EM-HMM tracking algorithm is naturally developed.Finally,two methods for initializing the EM-HMM algorithm are presented.Simulations show that all these are effective.关键词：multiple frequency line tracking;hidden Markov model;measurement probability;EM-HMM algorithm2|742|0更新时间：2025-07-16
- 摘要：By modifying the Lagrange multipliers,both equality and inequality constraints can be treated identically.Based on this idea,a new type of Lagrange nonlinear programming neural network is constructed,and the stability and convergence of the neural network are analyzed rigorously.Furthermore,the convergence condition of the network is relaxed by the approach of adding the penalty terms in the Lagrangian function.关键词：neural optimization;nonlinear programming;stability2|1157|3更新时间：2025-07-16
- 摘要：The paper sets up a class of dynamical fuzzy neural network systems,i.e.so-called (∨,T) fuzzy Hopfield networks based on L-T-norms.We prove the uniform stability of the system and the Lyapunov stability of the equilibrium points at first,then obtain a nontrivially attractive basin of the attractor so that the system has good fault-tolerance.Finally,simulation examples demonstrate our conclusions.2|814|4更新时间：2025-07-16
- 摘要：Firstly we analyse the properties of Strong Tracking Filter (STF) and Interacting Multiple Model Algorithm and find that STF is a parameter-adaptive algorithm and IMM is a model-adaptive algorithm.It means that they may be combined effectively.Secondly we propose a new method based on the Least-Squared Estimation to search for the fading factor in STF.After that,we design a Strong-Tracking Multiple Model Estimator (STMME) by combining the new STF with IMM.Finally,the simulations show that STMME greatly improves accuracy of velocity and acceleration compared with the conditional IMM when tracking the maneuvering target.And the computation burden increases only 6%.关键词：interacting multiple model algorithm;strong tracking filter;tracking maneuvering targets;fault detection3|866|6更新时间：2025-07-16
- 摘要：The need for an accurate simplification of surface meshes is becoming more and more urgent in many applications.It not only suits mulitiresolution/level-of-detailed representations and quick visualization,but also allows fast transmission of 3D models in network-based applications.Many different approaches and algorithms for mesh simplification have been proposed in the last few years.In this paper we present a new method,which partitions region by minimizing the maximum intercluster distance clustering algorithm,extracts features in every region and triangulates them by constrained delaunay triangulation.The results of examples illustrate our simplification method is effective.关键词：mesh simplification;surface segmentation;clustering;constrained triangulation2|1175|1更新时间：2025-07-16
- 摘要：This paper presents a new small target detection approach,which is based on multi-resolution distance map and background texture analysis. Through multi-scale wavelet decomposition of background textures and local energy calculation,feature vectors of each point are obtained.According to these feature vectors,a distance map corresponding to these features can be derived,and target detection is performed on this distance map.The histogram of distance map is employed for the automatic selection of threshold value.Experiments show that this approach can achieve quite satisfactory results.关键词：multi-resolution analysis;texture analysis;Target Detection3|1148|18更新时间：2025-07-16
- 摘要：A modified CHMM model is proposed to improve the robustness of the speaker recognition system.In the modified model,the difference of the speech feature parameter was the observed vectors of the state-transitions of CHMM.Because of the modification,the dependence between frames was used in the modified CHMM.By the fusion of different features,the robust feature will dominate under the noisy environment and the delicate feature will dominate under clean environment.The experiments indicate that the modified CHMM have effectively improved the robustness of the speaker recognition system.关键词：CHMM;Information fusion;robustness;speaker recognition2|1214|2更新时间：2025-07-16
- 摘要：We present in this paper a new two-level speech coding method-EMSVBR system,which consists of two subsystems.The core subsystem is based on hybrid coding technique and the subordinate subsystem is based on SBC technique that utilizes wavelet packet to construct SBC filter banks.Bit streams from this EMSVBR codec are layered and its bit rate is variable in wide ranges with the available capacity of networks or with the change of channel characteristic.This new codec demonstrates better performances than other algorithms if they operate at the same rate and therefore is suitable to be used in the cases of VOIP and VOA.关键词：speech signal processing;VBR coding;wavelet;bit allocation2|994|3更新时间：2025-07-16
- 摘要：Whole annealing genetic algorithm (WAGA) is a new genetic algorithm.It introduces annealing mechanism to selection operator and allows ancestor to take part in competition,which makes genetic algorithm more robust and efficient.In this paper,we apply WAGA to optimize soft morphological filters and study the performance of optimization algorithm with various noise percentages under MAE or MSE error criteria.The simulation experiments show that optimized soft morphological filters are highly improved and to suit for filtering of noisy images.关键词：soft morphological filters;genetic algorithms;optimization algorithms;image processing2|977|6更新时间：2025-07-16
- 摘要：Extracting a high-resolution image form a low-resolution image sequence with displacement solves the image super-resolution problem which is ill-posed to one-frame based interpolation.However it is indispensable to store all low-resolution images for traditional multi-frame based algorithm which can not be implemented in real-time.In this paper,we propose a multi-frame based recursive iteration algorithm which can achieve super-resolution result as good as traditional method but with less memory consumption.The proof of convergence and efficiency of this algorithm is also addressed.Computer simulation gives the result matching with theoretical analysis.关键词：resolution;image interpolation;image super resolution;recursive steep-descend iteration2|1104|3更新时间：2025-07-16
- 摘要：Based on analyzing the distributions of equipotential lines and capacitance sensitivity fields,this paper discusses the "soft fields" characteristics of electric field distributions for electrical capacitance tomography. The mechanism leading their inversions to nonlinear ill-posed problems is also studied.We employ a three-layer B-P neural network to reconstruct images. The input vector of the network is the preprocessed capacitance measurements and the output corresponds to the spatial image. Experimental results illustrate that this method is fast and accurate for image reconstruction.关键词：electrical capacitance tomography;FEM;nonlinear ill-posed problems;neural networks;image reconstruction2|1490|4更新时间：2025-07-16
- 摘要：The wavelet subband coefficients of images have highly non-Gaussian statistics that may be modeled with generalized Laplacian distributions,and Bayesian estimation is used to suppress noise.Because the wavelet transform is orthonormal,noises in wavelet coefficients of images corrupted by additive white Gaussian noises are also Gaussian and white,and performance of denoising via Bayesian estimator based on orthogonal wavelet transform is excellent.关键词：wavelet;Bayesian estimation;filtering2|1174|2更新时间：2025-07-16
- 摘要：This paper presents a robust digital image watemarking algorithm based on chaotic mapping.The watermark is preprocessed with a chaotic dynamical system and spread adaptively over the image spatial domain.Geometric calibration before watermark detection can be achieved through a preset known mask and the watermark multiple embedding.The experimental results demonstrate that the watermark is robust to general image processing techniques and geometric distortions.关键词：digital image watermarking;chaotic mapping;geometric calibration;robustness3|1222|3更新时间：2025-07-16
-
A New Wavelet Model for Identification of Discrete Chaotic Systems and Qualitative Analysis of Model
摘要：This paper develops a new multiresolution wavelet model for a special class of nonlinear system―discrete chaotic system.Based on the time-frequency localization properties of wavelet,an initial model is obtained.To obtain a parsimonious dynamical reconstruction,the structure selection algorithm based on match pursuit algorithm is implemented.Lowering-dimension decomposition method is proposed to solve the computational cost by the tensor product in high input dimensions.Simulation examples not only show the efficiency of the proposed method,but also compute dynamical invariants of model that is close to those of the original system.关键词：chaotic system;nonlinear-system identification;wavelet model;multiresolution analysis;lowering-dimension decomposition2|1004|3更新时间：2025-07-16 - 摘要：In this article,we study a chaotic global modeling method based on a set of polynomials defined on the attractor's invariant measure as its kernel.Through analysis of Hénon map data and observed data of ionospheric parameters,it is shown that if the underlying system is simple enough to be represented by low-order polynomials,this method can capture the dynamics of the system.Also,when the noise level is low,we can acquire the equation of the system.When noise level is higher or the system is more complex,we still can use the reconstructed model to perform one-step prediction in a fairly good fashion.关键词：chaos;global modeling;invariant measure of attractors;orthogonal polynomials;ionosphere2|1270|3更新时间：2025-07-16
- 摘要：The decomposition model and algorithm of high-accuracy recovery for weak signals are researched in this paper.The weak signals are contained in multivariable time series.Superposition of Logistic chaotic or non-chaotic series forms multivariable time series having cause of formation correlation.The numeral simulation analysis indicated this decomposition model and algorithm having capability of high-accuracy recovery for weak signals of Logistic chaotic or non-chaotic series.关键词：logistic chaotic series;superposition process;multivariable time series;weak signals;numeral simulation analysis2|951|3更新时间：2025-07-16
- 摘要：In this paper,we prove that oversampling the output of IIR channels can be transformed into a single input multiple output (SIMO) model,and the subchannels have the same AR coefficients.According to this model,a blind identification method is proposed based on subspace.The MA coefficients of the subchannels can be identified by using subspace decomposition.The AR coefficients can be resolved from Yule-Walker equation.关键词：oversampling;SIMO;subspace decomposition2|1002|2更新时间：2025-07-16
- 摘要：This paper discusses in detail the relations between the amplitude-frequency characteristic on the FIR filters with linear phase and the algorithm of the neural networks,proves the convergency theorem of the algorithm,and presents the design examples of the high-order multi-band-pass filters.The results show that the algorithm is efficient and excellent in the design field of the quasi-ideal multi-band-pass filters.关键词：neural-networks;high-order filters;multi-band-pass;optimum design2|1240|6更新时间：2025-07-16
- 摘要：A new soft-decision adaptive decoding algorithm for space-time trellis code (STTC) is proposed.With a bidirectional decoding (BD) structure,the proposed algorithm fully utilizes the preambles of adjacent TDMA downlink slots and produces high-quality soft-decision output.Besides,two-step Least-Mean-Square (TSLMS) algorithm is developed to implement simple but effective channel tracking in each unidirectional decoding.Computer simulation results are given to support our development.关键词：space-time trellis code (STTC);bidirectional decoding (BD);two-step Least-Mean-Square algorithm (TSLMS);soft-decision2|906|0更新时间：2025-07-16
- 摘要：We consider the design of a simple but powerful space-time trellis coding scheme.While delay diversity can be viewed as the combination of repetition code and delay element,GDDC(Generalized Delay Diversity Code) replaces the repetition code with some other simple block code in order to achieve higher coding gain.The necessary and sufficient condition for GDDCs to achieve full diversity advantage is given.It is proved that all the full diversity GDDCs on BPSK(Binary Phase-Shift Keying) constellation have the same coding gain.We have also thoroughly studied the coding gain of full diversity GDDCs on QPSK (Quadriphase Shift Keying) constellation and the optimal GDDC is given.The strategy can also be applied to the analysis of the coding gain of full diversity GDDCs defined on other constellations.关键词：space-time coding;delay diversity;performance analysis2|868|1更新时间：2025-07-16
- 摘要：In this paper,iterative joint demodulation and decoding of coded DPSK(differental phase shift keying)signals over an AWGN channel with unknown phase is presented.Firstly,the symbol-by-symbol maximum a posteriori probability (MAP) algorithm for an unknown phase channel is derived and it is shown that the MAP algorithm can be implemented efficiently using forward and backward recursions through the trellis.Secondly,the equivalent channel model approach is adopted and highlighted for subsequent systems to further process the channel MAP's output.Finally,the principle of iterative joint demodulation and decoding is introduced.The efficiency of the proposed algorithm is approved by many simulations.2|1277|1更新时间：2025-07-16
- 摘要：In the FGS (Fine Granularity Scalability) enhancement bitstream,the error resilient technique are not considered. when delivering video bitstream over error prone channel,the actual decoded bits are only a small part of received data due to the transmitted errors even if a great deal of enhancement bitstream is transmitted to the decoder. In this paper,a new architecture of enhancement layer bitstream with resynchronization markers and Header Extension Code (HEC) is proposed for efficient error localization and recovery. Experimental results show that,at the same channel conditions,the proposed enhancement bitstream is more robust than that of the original enhancement bitstream in MPEG-4 standard.关键词：fine granularity scalable video coding;enhancement layer bitstream;error resilience2|880|1更新时间：2025-07-16
- 摘要：An improved neural network detection method based on Logistic-map and BP (back propagation) network is proposed.It can be applied to phase acquisition of CDMA address code.The results of analysis and simulation both show that it can increase the detection probability and reduce the average acquisition time compared with conventional series phase acquisition method,and the performance of the whole mobile communication system can be improved effectively.关键词：chaos;neural network;address code;phase acquisition2|1037|0更新时间：2025-07-16
- 摘要：The research into the properties of neural networks is made in the framework of the dual linear programming theory.The variation range of a Hessian matrix as well as that of learning rate of the networks is confined by means of the eigenvalue.A Lipschith constant based formula for the algorithm's convergence is given.Simulation is given to illustrate that the theory is correct and the formula given is efficacious.关键词：dual linear programming;artificial neural networks;optimization;learning rate;simulation2|990|0更新时间：2025-07-16
- 摘要：The evolution of organisms takes place both in the genotype and the phenotype.At the phenotype level,the evolution is based on the laws of natural selection,while at the genotype level,it is a stochastic process of self-updating and self-optimization,and has a tendency of self-organizing to some degree.Based on this viewpoint of evolution,this paper proposes a new evolutionary algorithm and applies it to implement optimization.The algorithm considers not only natural selection of the phenotype,but also the evolution of genotype and the mappings between them.It has been demonstrated by simulation results that the algorithm is much better than other evolutionary algorithms based on typical frames in the aspects of convergence speed,robustness and global optimization ability.关键词：evolutionary algorithm;genetic algorithm;global optimization2|1003|5更新时间：2025-07-16
- 摘要：A design for Turbo trellis-coded modulation (TCM) systems is proposed with nearly half of the encoding delay caused by the conventional scheme.An introduction is also given to the associated constraint that the interleaver has to satisfy.An estimate of channel values in decoding is suggested so as to improve the accuracy of extrinsic information calculations and thus the decoding performance.The proposed iterative decoding algorithm,a natural extension of the standard decoding algorithm for TCM,is similar to the bit-wise one for binary turbo codes with the BPSK modulation.The bit error rate 10-5 of this scheme needs signal-to-noise ratios within 0.4dB from the Shannon limit 2.85dB for the throughput 2bits/s/Hz with the 8PSK modulation.关键词：Turbo codes;trellis-coded modulation;iterative decoding2|919|1更新时间：2025-07-16
- 摘要：A detailed study of loss-resilient codes based on random bipartite graphs are made in this paper.We propose the upper bound on the maximum tolerable loss fraction δ for which the decoding of the loss-resilient code with a given degree distribution pair is successfull.The algorithm to find the degree distribution pair of loss-resilient codes is presented by making a detailed analysis of this upper bound.This result shows that codes constructed from degree distribution pair obtained by the algorithm above should perform better than the degree distribution pair given in [2].Moreover,we prove loss-resilient code with the definite degree distribution pair can be both encoded and decoded successfully in linear time.关键词：loss-resilient code;random bipartite graph;encoding and decoding complexity;degree distribution pair algorithm2|817|1更新时间：2025-07-16
- 摘要：The elliptic curves cryptosystems (ECC) is the most popular public key cryptosystems at present.The hyperelliptic curve cryptosystems (HCC) is a generalization of ECC and the study of HCC has been paid more and more attention by people in recent years.This paper surveys the development and situation of HCC and presents some problems urgenthyneeded to be solved in the theory and implementation of HCC.关键词：hyperelliptic curve cryptsystems;jacobian;HCDLP2|2958|6更新时间：2025-07-16
- 摘要：In this paper,the soft-competition learning algorithms (SCLA) of RBF neural networks are designed.The main ideas of the algorithms are:firstly,membership functions are introduced into the training procedure of the center vectors of the Gaussian basis functions,and for each input sample,all center vectors are self-adaptively adjusted according to the values of the membership functions,in what degree the sample belongs to the classes that the center vectors represent;secondly,the reciprocal of the fuzzy factor of membership function are considered as the temperature of the simulated annealing algorithm,and increasingly adjusting method is used to the fuzzy factor during the learning procedure.SCLA are soft-competition schemes of the learning algorithms,in which the center vectors are trained based on the k-means algorithm,and can remedy the problems of the dead-node and the sensitivity to initial weight vectors that the latter algorithms have.The simulation results show that SCLA are efficient.关键词：radial basis function;soft-competition learning;fuzzy factor;simulated annealing2|1512|4更新时间：2025-07-16
- 摘要：Lifting scheme is generalized to perfectly reanstruct M-band(PR) FIR filter banks in this paper.Linear-phase property is studied in a special case.The design and optimization of filter banks can be commutated into a quadratic optimization problem.关键词：lifting scheme;M-band filter banks;perfect reconstruction;linear phase2|751|0更新时间：2025-07-16
- 摘要：The improved BP algorithm of multilayer neural network is disscused and a new method,weight balance algorithm is proposed to exert all weights' effect on the process of neural network training .Simulations indicate that it has a superior convergence rate and precision compared to the standard BP algorithm.The neural network structure which fits for this algorithm is also discussed.关键词：BP algorithm;neural network;weight balance algorithm2|1488|10更新时间：2025-07-16
- 摘要：Pseudo sparse-sequence has wide applications in information hiding technology.This paper presents several classes of pseudo sparse-sequences,named respectively product sequences,self-shrinking product sequences,self-expanding sequences,sparse sequences based on GF(2m),and sparse sequences based on power-residue-symbol.Their randomness is discussed,mainly on the least periods and linear complexities.关键词：stream cipher;information hiding;m-sequence;the least period;linear complexity2|852|1更新时间：2025-07-16
- 摘要：Fuzzy set methods have recently achieved a high degree of popularity due to its ability of representing and managing uncertainty,and become an effective method in information fusion fields.If the values of fuzzy measure,which can be got through optimization method,are appropriate,multiple classifier fusion method based on fuzzy integral performs better than the best classifier.In this paper,the Genetic Algorithm is adopted to search the optimized values of fuzzy measure,thus a satisfied fusion result better than that without GA can be obtained,as was verified in the experiment.关键词：Information fusion;fuzzy integral;genetic algorithm2|1103|10更新时间：2025-07-16
- 摘要：Hidden Markov Models are very successful in modeling the acoustic behavior of speech.They may be classified into two groups,continuous HMMs (CHMMs) and discrete HMMs (DHMMs),according to the output probability distribution.Traditional training methods such as Baum-Welch algorithm are noted for the rapid convergence.However,these methods are hill-climbing based algorithms and they just lead to locally optimal solutions,which might deteriorate the recognition rate.For CHMMs,a segmental k-means method has been developed to get reliable initial estimate and thus guaranteed the rapid and proper convergence.For DHMMs,this offers little help and the final solution is a locally optimal solution.While one outstanding character of evolutionary computation is global search,it can converge to a globally optimal solution or at least a sub-optimal solution.In this paper evolutionary computation is applied to training DHMMs.A hybrid training method that combines the traditional method and evolutionary computation is proposed.Experimental results show that the proposed method has both qualities of global search and rapid convergence and the resulting models are superior to those obtained with traditional methods or simple evolutionary computation and eventually contribute to the increase of recognition rate.关键词：HMMs;evolutionary computation;cpeech recognition2|1077|2更新时间：2025-07-16
- 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