最新刊期
卷 ， 期 11 ， 1999
- 摘要：Time series prediction can be applied to a lot of fields such as financial analysis etc.Recently,there are growing interests in neural network based time series predictors.However,neural network based time series predictor often gives invalid predictions.In this paper,the probability of invalid prediction given by neural network based is analyzed firstly.And then a new Regularized Learning algorithm is proposed for Neural Network based time series Predictors （RLNNP）.After sufficient training,neural network based time series predictor with RLNNP algorithm will give a valid prediction for time series.关键词：neural network;time series prediction;regularized learning algorithm2|341|21更新时间：2025-12-08
- 摘要：This paper proposes a new minimum shortening channel square error （MSCSE） criterion for optimizing the time domain equalizer in orthogonal frequency division multiplexing （OFDM） systems.From the new criterion,a very practical algorithm is derived.The paper also proves that the conventional MMSE criterion will be converted into MSCSE criterion on certain conditions.The performance comparisons between different algorithms are carried out.关键词：time domain equalization;Orthogonal frequency division multiplexing;digital communication2|138|2更新时间：2025-12-08
- 摘要：This paper discusses some problems of intelligent control engineering （ICE）.Firstly,two key techniques of ICE,computation techniques and integration techniques,are introduced.Furthermore,based on the viewpoints of cognitive science,a cognitive frame of intelligent control in which human machine combination and hybrid techniques integration can be carried out,is proposed in the form of intelligent information processing.Some methods are also given to realize composed parts of the frame.A designed example is provided subsequently.关键词：intelligent control engineering;computation technique;integration technique2|173|2更新时间：2025-12-08
- 摘要：Microstructure of P + Si 0 65 Ge 0 35 /p Si heterojunction internal photoemission （HIP） infrared detector has been observed by using localization cross section transmission electron microscopy.The photo sensitive area in the detector consists of 6 P + Si 0 65 Ge 0 35 layers and 5 UD Si layers,which are smooth and have thickness of 6nm and 32nm,respectively.A stress field exists on the interface between Si 0 65 Ge 0 35 and UD Si layers,but the crystal defects have not been found in this area.Both Si 0 65 Ge 0 35 and UD Si layers on amorphous SiO 2 step consist of polycrystals and are in the shape of waves.There are defect regions in the edger of photo sensitive area.The region is less than 120nm in width.The crystal defects are stacking faults and microtwins.关键词：infrared detector;heterojunction semiconductor device;microstructure of semiconductor device;transmission electron microscopy2|46|0更新时间：2025-12-08
- 摘要：Based on the feature analysis of fault diagnosis methods now available and their disadvantages used in multi sensors’ fault diagnosis and classification,a novel fuzzy logic system （FLS） approach for the fault diagnosis and classification of multi sensors,which combines a network form multi output FLS with a fault diagnosis and classification method of sensors using single output adaptive fuzzy logic sytem,is proposed in this paper.The validity of the proposed approach is verified by simulation.关键词：fault diagnosis;classifcation;multi sensors system;fuzzy logic system2|133|1更新时间：2025-12-08
- 摘要：Wavelength conversion due to four wave mixing（FWM） in a semiconductor optical amplifier （SOA） is theoretically analyzed in this paper.An analytic expression is developed describing the relationship between conversion efficiency and the characteristic parameters of SOA and input pump power.The numerical simulation results show that the wavelength conversion efficiency is influenced by the input pump power and saturation power of SOA.The theoretical evidences are given for the design of wavelength converter.关键词：wavelength conversion;four wave mixing;semiconductor optical amplifier2|235|5更新时间：2025-12-08
- 摘要：Two new universal active current mode filters are presented.One is with three inputs and one output by using two current conveyors（CCII）,at which various filtering functions such as second order low pass,high pass,band pass,notch and all pass filtering are obtained by selecting different input signal combinations.The other is with single input and three outputs by using two current conveyors and multiple outputs Operational Transconductance Amplifier（OTA）.The proposed circuits are with simple structure,seperately adjustable centre frequency ω 0 and the quality factor Q ,all grounded passive components and low passive sensitivities.Theoretical analysis and computer simulation indicate that the proposed circuits are effective.关键词：current mode circuit;current conveyor（CCII）;operational transconductance amplifier（OTA）;active filter2|235|49更新时间：2025-12-08
- 摘要：A tool for dependability evaluation of fault tolerant systems:HFI 2 fault injector is introduced.The function and structure of this tool is described,and some key techniques are discussed.The effectiveness of this tool is validated by experimentalresults.关键词：fault injection;fault tolerant;dependability evaluation;coverage2|349|42更新时间：2025-12-08
- 摘要：In this paper we study how the Fractal Wavelet transforms perform in image compression and give a framework for this class of image coding method.We also treat“horizontal”、“vertical”and“diagonal”scaling factors of quadtrees independently according to the statistic analysis of test image orientation property and get a result in a marked improvement in accuracy with respect to quantization for image compression.关键词：fractal;wavelet transform;orientation property;IFSW transform;image compression2|141|13更新时间：2025-12-08
- 摘要：High temperature Al 0 3 Ga 0 22 In 0 48 P/GaAs heterojunction bipolar transistors （HBTs） were fabricated,and their DC electric characteristics were studied in detail from 300K up to 623K.Experimental results show that Al 0 3 Ga 0 22 In 0 48 P/GaAs HBT possess excellent high temperature properties.Over the temperature range from 300K to 623K,current gain changes less than 10% from its average value.At ambient temperature higher than 623K,HBT failed.Analysis has shown that the short circuited BC junction is the direct cause of HTB’s failure.Further research works need to be done in order to fully understand BC junction’s failure mechanism.关键词：heterojunction bipolar transistor （HBT）;high temperature;DC electric characteristics2|23|0更新时间：2025-12-08
- 摘要：In this paper,the Coherent Measurement techniques are used to derive equations of fundamental waves above or below Nyquist frequencies.And then the high speed analog to digital converter has been tested using the LTX’s SYCHRO Ⅱ measurement system.The Full Power Bandwidth （FPBW） and some ADC dynamic characteristics have been examined.The results are in good agreement with the ADC’s parameters given by AD Corporation.关键词：full power bandwidth;high speed analog to digital converter;coherent measurement;digital signal processing2|174|8更新时间：2025-12-08
- 摘要：A current mode novel multifunctional WTA network with smart structure is put forward in this paper.This network is reconfigurable with a high precision.It can also realize both absolute rejecting fuction and relative rejecting function.The weighted sum of all the input currents is regarded as a relative threshold in the first kind of relative rejecting when this network is configured in 2 WTA and so it has some adaptibility.If the difference between the first maximum and the second maximum input currents is less than a given value or a certain percentage of the first maximum input current,then a relative rejecting signal is given by the second kind of network.In this case,the network is usually used as a general 1 WTA.These two kinds of rejecting play an important role in the field of pattern recognition.HSPICE simulation of this network gives a good performance.关键词：WTA network;absolute rejecting;relative rejecting;current mode circuit2|40|3更新时间：2025-12-08
- 摘要：Based on a new approach to solve the Electrocardiograph （ECG） inverse problems,we study the recognition of the myocardial infarction （MI） parameters.The relations between the MI parameters （MI sites and sizes） and the body surface potential mappings （BSPM） are studied in detail using the heart simulation model.With the use of the BSPM features that are sensitive to the MI sites and sizes,mathematical models of the optimization system are structured.The experimental results indicate that the proposed approach in this paper can recognize efficiently the MI sites and sizes and the recognition accuracy to the MI sites is a myocardial segment in the heart model and to the MI sizes larger than 85%.关键词：electrocardiograph inverse problem;myocardial infarction;heart simulation model;body surface potential mapping3|50|3更新时间：2025-12-08
- 摘要：Debugging distributed applications is commonly thought to be very difficult.and one of the problems is that distributed applications are inherently more complex than sequential ones.To manage this problem,the method of abstracting events is proposed,which lets the users grasp various aspects of behaviors of a distributed application.Abstract event time stamp is used to decide the `happened before’ relation among abstract events,which plays a very important role during abstracting and debugging distributed applications.We provide a new incompleted algorithm of timestamping abstract event based on Lamport logical time,which needs less storage and costs less time than others.Proof of the correctness of the algorithm is also given.关键词：distributed system;debugging;abstraction;timestamp2|58|5更新时间：2025-12-08
- 摘要：The optimization based and hierarchical approach is presented,which is used to synthesize D/A converters automatically.In this method the specifications for building blocks are determined from customer performance requirements firstly,which are then used to drive synthesis of building blocks.The optimization at both levels is performed to enable optimumal results.Besides,synthesis results are evaluated in a bottom up way.关键词：analog circuit;CAD;D/A converter2|40|1更新时间：2025-12-08
- 摘要：A new approach for call admission control using fuzzy logic based on neural networks is proposed.This approach can select preliminary parameters properly,have self learning ability by combining the benefits of neural networks and fuzzy logic.Simulation results show that the approach can use less training data,has higher decision accuracy,and saves a large smount of training time,so it is superior in comparison with analytical methods and neural networks methods.关键词：CAC;fuzzy logic;neural network;effective bandwidth2|34|0更新时间：2025-12-08
- 摘要：A new very low bitrate video compress algorithm based on Vector Quantization is presented in this paper.The proposed approach involves quadtree segmentation,classified Vector Quantization and variable size motion compensation simultaneously.Simulation result shows that this algorithm achieves good subjective image quality while gets high compression ratio.Considering the simple structure of the decoder,here we prsent an encoder architecture which is suitable for VLSI implementation.In this architecture,the motion estimation can share the same Process Array with Vector Quantization in order to reduce the chip area.关键词：very low bitrate video coding;vector quantiztiong;VLSI architecture2|35|1更新时间：2025-12-08
- 摘要：Morphological filters perform nonlinear signal transformations that locally modify geometric features of signals.It has been shown that morphological filters can be efficiently used to filter noise in binary images and gray scale images.In this paper,color morphological operations,based on the RGB （red green blue） color space,are introduced,which extend theory of mathematical morphology to color image processing.The most fundamental color morphological filters,color opening and color closing,are proposed and used to construct more powerful color morphological filters.Analyses are developed.Experimental results show that color morphological filters are very suitable and powerful to eliminate noise of color images without blurring image and producing new color.They perform better than vector median filter.关键词：mathematical morphology;RGB color space;vector norm;color morphological filter;vector median filter4|132|26更新时间：2025-12-08
- 摘要：In this paper,we present an algorithm for routing the terminal stacked channel with four or five layers.This algorithm’s object is to route a channel with the minimum channel row number and the over cell routing is used.In the channel routing,the VCM is introduced to determine the routing sequence of the nets.The algorithm can always guarantee the success of the channel routing,and it runs fast and performs near optimally （the result is optimal or one row more than the optimal result） on all examples we implemented.关键词：channel routing;feed through;over cell routing;VCM;virtual terminal;backtrack2|32|0更新时间：2025-12-08
- 摘要：Starting with the intrinsic current voltage properties of Capacitively coupled Single Electron Transistor （C SET）,the intrinsic current voltage properties of each form of the C SET active loads are studied by using a quasi classical model of the single electron tunneling effect,and its small signal equivalent circuit is also discussed.The results are significant for the design of the single electron analog circuits.关键词：capacitively coupled single electron transistor;active load;AC small signal equivalent circuit2|62|6更新时间：2025-12-08
- 摘要：The strain distribution of the twin isles structure polysillcon pressure sensor has been simulated by computer with the finite element method.According to the law of the strain distribution,polysilicon piezoresistor has been optimally designed.Moreover,the disadvantage which piezoresistor coefficient of polysilicon is lower than monocrystalline silicon is remedied.A new twin isles polysilicon pressure sensor has been designed and fabricated.The sensor has high sensitivity and elevated operation temperature.The experimental results in accord with the simulation.关键词：twin isles structure;finite element method;polysilicon;sensitivity2|125|8更新时间：2025-12-08
- 摘要：Some differences in property between signal and noise are used to restore the original signal from the polluted signal.This process depends on accurate description of these properties.The filtering feature of wavelet transformation is analyzed in the paper and the threshold filtering algorithm is studied.The sernse of the threshold for filtering is explicated and the role to get the threshold is given.The mathematical expression of threshold is given and the adaptive threshold filtering algorithm is constructed.关键词：wavelet transformation;filtering;threshold2|279|63更新时间：2025-12-08
- 摘要：It is very important in many study fields to estimate multi parameters of multi signals in broad band,such as recognizing of unknown emitting sources,active canceling and so on.The problems of estimating signal frequency along with DOA and DOA along with polarizations have been studied.But it hasn’t been reported the method of joint estimating of frequency,DOA and polarizations yet.A new method that joint frequency,two dimensional arrival angles and the polarizations estimation of incoming narrow band signals use the ESPRIT algorithm and a square array of crossed dipoles is proposed.关键词：array;algorithm;frequency;two dimensional arrival angles and the polarizations estimation3|274|66更新时间：2025-12-08
- 摘要：In the radar field,performance of pulse compression system can be improved through Non Linear Frequency Modulation （NLFM）.On the basis of our model of frequency modulation function,a method,which can generate NLFM signal that mects the requirement of pulse compression system,is presented in this paper.An example of designing NLFM signal via the method is given.Computing result demonstrates that the method can design quickly and flexibly the desired signal.关键词：genetic algorithms;non linear frequency modulation;frequency modulation function4|301|16更新时间：2025-12-08
- 摘要：An one dimensional fluid model for discharge occurring in AC PDP cell is developed,and the multi pulse discharge characteristics of AC PDP cell is studied by numerical analysis in this paper.This model is based on the dual fluid description of electron and ion transfer,include the particles continuity equations and momentum transfer equations,coupled with the Poisson equation and circuit equation.In calculation,the filled gas is assumed as helium and in the local equilibrium state,in which the reacting rate of every kind particles is assumed to be a function of E/P and worked out by solving the stable Boltzmann equation in He.In this paper,the behavior of multi pulse discharge under 50kHz square waveform driving voltage occurring in AC PDP cell is simulated,and the distributions in space and variation as time of some parameters during discharge are calculated.These parameters include the voltage,discharge current and the density of kinds of particles and potential etc.关键词：AC plasma display panel （PDP） cell;discharge fluid model;multi pulse discharge characteristics;numerical simulation2|57|2更新时间：2025-12-08
- 摘要：In the light of a modified two dimensonal Poisson equation,an analytical model for the threshold voltage of deep submicron MOSFETs is developed and,by introducting a continuous channel charge characteristics and properly including major small geometry and high field effects such as carrier velocity saturation,DIBL,field dependent mobility,inversion layer capactitance and channel length modulation etc.,an accurate,continuous and scaled down drain current model is obtained.The model’s output fits both the experimental data of Huajing’s sample and MINIMOS simulation well in all operation regions,and is avalable for VLSI simulation applications.关键词：short channel;deep submicron;MOSFET;threshold voltage;drain current;analytical physical model2|176|1更新时间：2025-12-08
- 摘要：This paper proposes a delay estimation method for different interconnect tree topologies.Using this method,we can incrementally compute transfer function for general interconnection trees and construct a two pole transfer function to approximate the actual transfer function by coefficients matching,not the moment computation,then derive a two pole model for delay estimation.The delay estimates using our method are very close to spice computed delays.And the trade off between accuracy and efficiency can be obtained.It is useful for delay calculation of interconnection in the performance driven synthesis and layout of VLSI routing topologies.关键词：interconnection in VLSI;delay estimation;two pole model2|59|2更新时间：2025-12-08
- 摘要：High performance parallel computer are increasingly in demand in the areas of the large scale scientific computations,such as in oil reservoir.In this paper,a large scale parallel numerical simulation of reservoir problem that has more than 728,000 grid points is solved by applying parallel theory and technology based on distributed memory parallel systems with several shared main memory multiprocessors joined by network.This parallelism software improves computation capacity and speeds considerably in the aspect of numerical simulation of reservoir.关键词：parallel computation;large scale numerical simulation of reservoir;distributed parallel;load balarcing2|140|6更新时间：2025-12-08
- 摘要：In this paper,we present typical application—oriented computer network models,analyse the reliability and availability of these models,and discuss a practical example.Finally,it is pointed out how the appropriate model is designed and selected for a practical application.关键词：computer network;reliability;availability2|92|6更新时间：2025-12-08
- 摘要：Reducing for knowledge is an important step in KDD.In theory of rough sets,the principle and methods of the reducing for attribute are specially discussed.Many scholars work over arithmetic of the reduction,but the researches are almost for static data.However databases are dynamic.The paper provides the arithmetic,ASRAI,to get the sets of the new smallest attribute reductions from to the old sets ofter a record is added to the database.The arithmetic is proved to be right and the time complexity of the arithmetic is analyzed.Then the improved arithmetic,IASRAI,is provided.关键词：KDD,粗糙集合;增量式属性约简2|377|113更新时间：2025-12-08
- 摘要：A learning algorithm of the fuzzy counterpropagation （FCP） neural network consists of two parts:updating the weight vectors from the input layer to the competitive layer,and the weight vectors from the competitive layer to the output layer.To use the clustering method and the gradient method to respectively update the former vectors,two learning algorithms of FCP network are proposed.The properties of the algorithms are analyzed,and some important theories are achieved.Finally,to apply the algorithms to a time series prediction problem based on the Mackey Glass equation,the experimental results show that the latter algorithm outperforms the former algorithm in learning precision and generalization ability,but learning speed of the former algorithm is faster than of the latter algorithm.关键词：对向传播神经网络;模糊度因子;时间序列预测2|117|4更新时间：2025-12-08
- 摘要：Aiming at the problem of dynamically gain imbalance in cascaded amplification WDM system,a series of dynamically gain equalized Erbium doped fiber power booster and inline amplifier were designed and built,in which the population inversion was controlled.The equalisation transmission was demonstrated by using this kind of EDFA.关键词：erbium doped fiber amplifier;dynamic gain equalization2|35|0更新时间：2025-12-08
- 摘要：Usually drift signal has a first order continuity while sensing variable has a discrete change in some sensors.According to the different characteristics between drift signal and discrete change sensing signal,the difference of output samples are used to detach sensing signal from sensing output mixed with drift signal.It cancels the drift without any compensation devices.关键词：drift;sensor;data processing2|104|5更新时间：2025-12-08
- 摘要：A specific VLSI design with advanced pipeline architecture for fuzzy micro controller is presented in this paper.It uses way of single instruction flow and multi dataflow（SIMD）,and adopts some advanced structures,including;superscalar pipeline,precontract table,etc.so it has better performances that cycles per fuzzy instruction （CPI） is 0 5.This circuit is simulated by the hardware description language （VHDL）.Synthesis results show that the IC chip contains about 20k unit area （include 2k RAM）,and the average fuzzy logical inference per second（FLIPS） can reach 60M rules/s when 1 5μm CMOS process technology is used.关键词：pipeline;fuzzy micro controller（FMC）;VLSI;VHDL synthesis2|39|0更新时间：2025-12-08
- 摘要：After the technical restructure implemented in August 1997 in Changchun Artificial Satellite Observatory,the range precision and stability of the satellite laser ranging system have been greatly improved.A high quantum efficiency,low time walk,compensated,low work voltage and fast response time single photon avalanche diode was applied to the system,besides the establishment of a portable calibration system and three new near ground targets.Now,the system produces subcentimeter single shot precision and precision in mm level for normal point data.The long term system stability is improved from 4 cm to better than 2 cm,the range bias is less than 6 cm,and the time bias is less than 50 μs.关键词：satellite laser ranging （SLR）;single photon avalanche diode （SPAD）;portable calibration system （PCS）;range precision2|206|3更新时间：2025-12-08
- 摘要：The Differential Quadrature Method（DQM） is proposed in this paper for analysis of the transient response of the interconnects for high speed IC systems.DQM is a direct numerical technique,based on representing a derivative operator by the function values at discrete points,so the partial differential equations are reduced to the ordinary differential equations.DQM can be effciently used for transient analysis of the interconnects of high speed IC,and the computing efficiency of which is better than other numerical methods.关键词：differential quadrature method;high speed VLSI;interconnect;transient analysis2|75|10更新时间：2025-12-08
- 摘要：Based on the Fuzzy Neural Networks Sliding Mode Controller （FNNSMC）,we present a direct adaptive control scheme for a class of nonlinear systems in the presence of disturbances and model uncertainties.The FNNSMC and the Sliding Mode Controller with Dead Zone （SMCDZ） are incorporated to achieve the adaptive control by the smooth transformation.This methodology enhances the system’s robustness and can eliminate the high frequency chattering efficiently.Since the adaptive control scheme is used,the system obtains good tracking behavior.The simulation results demonstrate the advantages of the scheme.关键词：nonlinear systems;fuzzy neural networks;sliding mode control;direct adaptive control2|211|5更新时间：2025-12-08
- 摘要：Delay testing is important for high speed ICs.The main difficulty for delay testing comes from the huge number of paths and the large percentage of delay untestable paths.This paper presents an approach to delay testing with duplicating variable observation points,which provides a high path delay fault coverage by testing a small number of paths.The penalty that we should pay is a variable double observation at the output to identify the output transition time.This approach simplifies delay testing greatly in that the number of sample paths is linear to the number of gates in the circuit under test,despite of exponential growth in the number of paths.关键词：delay testing;linearly independent;observation2|39|7更新时间：2025-12-08
- 摘要：MCM is a new technology of ICs.In the design steps of MCM,partitioning is very important.Based on the improvement to Kohonen self organizing neural nerwork,the neural network approach for the timing driven system partitioning on MCM is presented.In the algorithm,the total routing cost between the chips and the circle time are both minimized,while satisfying area and timing constraints.The similarity of the directly connected cells is described by the connection weight between the cells and the internal delay of the combinatorial cell,and fuzzy similarity computing is applied to describing the similarity of indirectly connected cells.Each cell is mapped noto a two dimensional plane,and corresponds to one or a few neurons.The neural network is learned from a sequence of movements of cells,allows corresponding cells to cooperate during the learning process.By the self organizing learning process,the cells that have great similarity with each other will move as closely as possible.关键词：self organizing neural network;MCM;system partitioning2|31|0更新时间：2025-12-08
- 摘要：The influence of the decreasing of the voltage along the grid line on the design of front contact was analyzed.The relation between the power loss and the width of the finger and grid was discussed and the method of obtaining the ideal finger and grid from the preliminary design has been shown in this paper.关键词：solar cell;grid line design;power loss2|506|11更新时间：2025-12-08
-
Measurement of Radiation Induced Interface Traps Using the Subthreshold I-V Characteristic of MOSFET
摘要：This article studies the measurement of radiation induced interface traps using the subthreshold I V characteristic of MOSFET.The radiation induced interface traps which were related to total dose and dose rate were analyzed and discussed.关键词：I-V characteristic;radiation;interface traps2|64|1更新时间：2025-12-08 - 摘要：This paper describes a generic behavioral model pvreg which is implemented with AHDL.The model covers most linear regulator ICs in the world and has such function as dynamic thermal modeling.Behavioral models of various kinds of linear serial voltage regulator can be generated easily with pvreg.An example is given at the end of the paper to demonstrate its versatility in modeling ICs.关键词：AHDL;behavioral model;linear voltage regulator2|29|1更新时间：2025-12-08
- 摘要：Enlightened by the loop shaping theory,a new control algorithm based on closed loop gain shaping is presented through survey of the singular value curve of H ∞ S/T complex sensitivity as well as the correlativity of S and T .The algorithm shapes the closed loop gain without selecting weighting functions.Its core is to determine the final shape of T that is the transfer function of the closed loop system,and the robust controller is directly constructed using parameters having engineering sense.The algorithm guarantees fundamentally that the designed controller has fine control performance and robustness.Based on this algorithm a kind of new robust control algorithm for ships is given.The new algorithm has the advantages of simple design procedure and obvious physical sense.关键词：closed loop gain shaping;correlativity of S and T;autopilot for ships;robust control2|287|51更新时间：2025-12-08
- 摘要：Mechanically linearly chirped tunable 10cm uniform fiber grating for dispersion compensation has been described in this paper.In the mean time,the 2 5Gb/s 103km traditional fiber transmission experiments and dispersion compensation has been implemented.Experimental results show that,this method can adjust the grating reflective center wavelength as well as its chirp parameter.关键词：chirped grating;dispersion compensation2|62|1更新时间：2025-12-08
- 摘要：This paper proposes a protocol that enables communication over public channel to be authenticated by the partially secret common information between the two parties,and makes privacy amplification secure against active adversaries with some probability.At last,the relationship between the block size and the success probability of active attack is analyzed and a lower bound on size of bits of the common string needed to implement privacy amplification is given.关键词：privacy amplification;message authentication;unconditionally security;secret key agreement2|89|1更新时间：2025-12-08
- 摘要：The new concept of Gas Pendulum is put forward in this paper.Buoyancy lift and the pendulous characteristics of gas flow,the analogy between gas pendulum and solid pendulum and solid pendulum,and the gas pendulum acceleration sensor are also discussed.关键词：gas pendulum;solid pendulum;liquid pendulum;acceleration sensor2|76|22更新时间：2025-12-08
- 摘要：A VLSI implementation oriented off line handwritten digit classification system based on cellular neural network and two layer perceptron is presented in this paper.The cellular neural network is used to extract the connected component feature of normalized 20×20 handwritten digit image.These features are fed into the two layer perceptron after proper compression in time sharing mode.The two layer perceptron has a structure of 80×20×10 with local connection and it mainly consists of a weight programmable NPU array,switched current integrator and current comparators.This classification system has a smart structure and it can be directly mapped into current mode circuits whose technology is compatible with standard digital CMOS VLSI.关键词：cellular neural network;multi layer perceptron;handwritten digit recognition;switched current;VLSI2|102|3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