Product
Patsnap Eureka
For R&D, Patsnap Eureka makes reading and utilizing patents & technical documents easy.
Patsnap Eureka AIR
Designed for self-driven R&D workflows. Generate viable solutions, solve complex R&D challenges, empower your innovation with AI.
Patsnap Eureka Materials
Designed for material experts only. Revolutionize your material R&D, from search, analyze, to developing new materials.
Feature
TechResearch
Generate reliable direction feasibility study reports for your R&D in just a few steps.
TechSeek
Discover and master advanced knowledge NOW. Basics, ideas, possibilities, all at once.
TechMind
As an expert in R&D Theories, TechMind can generates customized viable solutions instantly.
TechRisk
Analyze your overall solution with one click, know your potential R&D risks in advance.
TechMonitor
Get weekly tech updates, stay abreast of the latest tech innovations and key insights.
Patents
Literature
Hiro is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
47 results about "Channel quantization" patented technology
Efficacy Topic
Property
Owner
Technical Advancement
Application Domain
Technology Topic
Technology Field Word
Patent Country/Region
Patent Type
Patent Status
Application Year
Inventor
Multi-user MIMO system precoding method and apparatus based on limited feedback
InactiveCN101369871AImprove performanceReduce quantization errorTransmitter/receiver shaping networksError prevention/detection by diversity receptionSystem capacityEngineering
The invention discloses a channel quantization method of a multiuser MIMO system based on a limited feedback and a device thereof belonging to the wireless transmission technology field. The method comprises the following steps: firstly choosing a code word having a minimal included angle with a codebook from the codebook according to the channel state information, and calculating the quantized error; secondly choosing a code word vector having a minimal included angle with the quantized error from the codebook; projecting the channel vector to the space formed by the chosen two code word vectors and calculating to obtain a compound proportion factor; quantizing the compound proportion factor using a pre-generated quantization step set. A receiving terminal device comprises a channel condition information acquisition unit, a channel vector quantization unit and a quantization information feedback unit. The transmission terminal device comprises a feedback information reconstruction unit and a precoding unit. The channel quantization method of the invention can reduce the quantization error, improve the system capacity performance and have advantages of small used codebook and low complexity when the speed and capacity are the same.
Owner:SHANGHAI JIAO TONG UNIV
Method and device for quantizing time and frequency channel and corresponding mobile communication terminal and corresponding mobile communication system
ActiveCN101540746AExact matchReduce distractionsMulti-frequency code systemsOrthogonal multiplexResource blockGranularity
The invention discloses a method and a device for quantizing a time and frequency channel and a corresponding mobile communication terminal and a corresponding mobile communication system. The mobile communication terminal of the invention comprises a time and frequency channel quantization device, and the method for implementing the device comprises the following steps of: A, measuring the channel state of the mobile communication terminal to obtain channel state vectors in the symbols of subcarrier waves in a physical resource block (PRB) of an orthogonal frequency-division multiplexing (OFDM) system; B, estimating related matrixes of the channel state vectors; C, performing the characteristic value decomposition of the related matrixes to obtain characteristic vectors in the physical resource block; and D, quantizing the characteristic vectors in a preset codebook. The mobile communication terminal informs a base station of code vectors obtained through the quantization; and the base station performs weighing according to the received code vectors to obtain a transmission signal to transmit the transmission signal. The method can reduce the influence of the granularity in time-frequency domains, match channels more accurately, reduce interference among multiple users and improve MU-MIMO capacity.
Owner:DATANG MOBILE COMM EQUIP CO LTD
Method and device for generating precoding matrix in wireless communication system
InactiveUS20130202059A1Improve performanceSpatial transmit diversityModulated-carrier systemsCommunications systemDirection information
Provided is a method and a device for generating a precoding matrix in a wireless communication system. The method for generating the precoding matrix comprises a step of receiving from a terminal feedback including information on the error angle θ of channel direction information (CDI), which indicates the direction of a channel direction vector, and a step of generating the precoding matrix based on the information on the error angle θ of the CDI, wherein the error angle θ of the CDI indicates the angle between the channel direction vector and a quantized channel direction vector, and includes at least one of a channel quantization error component, a channel feedback delay error component, and a channel estimation error component.
Owner:LG ELECTRONICS INC +1
Low-bit quantization method of depth separable convolution structure
ActiveCN110659734AImprove accuracyReduce workloadNeural architecturesNeural learning methodsMoving averageLabeled data
The invention discloses a low-bit quantization method for a depth separable convolution structure. The method comprises the following steps: selecting a quantization coefficient to uniformly quantizethe weight of a trained depth separable convolution neural network and feature map data; performing channel-by-channel quantification on the weight in the trained deep separable convolutional neural network; quantifying the depth separable convolutional neural network layer by layer by utilizing the feature map data; performing forward operation in the trained depth separable convolutional neuralnetwork based on the training set, and updating moving average parameters in the batch normalization layer; and fusing the updated moving average parameters and learnable parameters in the batch normalization layer into a network weight quantization coefficient and offset to realize low-bit quantization of the depth separable convolution structure. The low-bit quantization method of the depth separable convolution structure provided by the invention does not need any labeled data participation, is simple in calculation, and improves the accuracy after low-bit quantization compared with a traditional method.
Owner:INST OF SEMICONDUCTORS - CHINESE ACAD OF SCI
Method and system for an efficient channel quantization method for MIMO pre-coding systems
InactiveUS20080165044A1Electric signal transmission systemsEqual length code transmitterChannel state informationUnitary matrix
Aspects of a method and system for an efficient channel quantization method for MIMO pre-coding systems may include quantizing channel state information in a MIMO pre-coding system onto a codebook, which comprises one or more unitary matrices, using a cost function; selecting a quantization function that enables the quantizing based on a number of receive antennas and a number of transmit antennas associated with the MIMO pre-coding system; and selecting the cost function from a set of cost functions based on the number of receive antennas and the number of transmit antennas. The channel state information may be a matrix Ĥ. The set of cost functions may comprise function f(A), which may be defined by the following relationships:f(A)=∑i=1N∑j=1,j≠iNai,j2and / orf(A)=∑i=1N∑j=1,j≠iNai,j2-∑i=1Nai,iwhere A is a matrix of size N by N and aij is element (i,j) of matrix A.
Owner:AVAGO TECH INT SALES PTE LTD
Interference alignment method in heterogeneous cellular network
InactiveCN104579444AImprove performanceSuppress co-channel interferenceSpatial transmit diversityModulated-carrier systemsChordal distanceBit allocation
The invention provides an interference alignment method in a heterogeneous cellular network. With the adoption of the interference alignment method in the heterogeneous cellular network provided by the invention, same frequency interference can be effectively inhibited in terms of interference overlapping, and accordingly, the performance of a user is improved. The interference alignment method overcomes the defect that in a conventional technique for coordinately processing inter-cell interference with a resource orthogonal division method, each user occupies a few channel resources; in order to solve the problem that influence of non-ideal CSI (customer satisfaction index) on the performance of a system is not taken into consideration in a conventional heterogeneous cellular network interference alignment technique, the interference alignment method performs interference alignment design in a limiting feedback system, takes influence of non-ideal CSI on the performance of the system into consideration, performs bit allocation at the aim of minimization of average speed loss and performs channel quantization under the principle of minimization of chordal distance.
Owner:WUXI BUPT SENSING TECH & IND ACADEMY +1
Transmission beamforming method and apparatus in an orthogonal frequency division multiplexing-based MIMO wireless system
ActiveUS20120321017A1Reduce errorsReduce overheadSecret communicationRadio transmissionCurrent channelCommunications system
The present invention relates to a transmission beamforming method and apparatus in an orthogonal frequency division multiplexing (OFDM)-based MIMO wireless system. In particular, the present invention relates to a transmission beamforming method in an OFDM-based MIMO wireless system and an apparatus therefor, wherein the transmission beamforming method comprises: estimating a channel from a received pilot signal and obtaining time for the estimating channel and 2D channel correlation information of a frequency band; obtaining estimated channel information based on the channel correlation information; obtaining differential information that represents a difference between the current channel and the estimated channel and quantizing the differential information; and generating a transmission beam weight by using a predefined code book from the quantized differential information. The present invention may also provide a method and apparatus for reducing the channel quantization error and feedback load by employing the OFDM-based MIMO in a wireless communication system using a transmission beamforming scheme, wherein a receiver estimates current channel information by using time and 2D channel correlation information of a frequency based on the previous channel information, compares the estimated current channel information with an actual channel and then feeds back the differential channel information, i.e., the difference between the estimated channel and the actual channel, to a transmitter.
Owner:SEOUL NAT UNIV R&DB FOUND
Method and system for an efficient channel quantization method for MIMO pre-coding systems
A method and system for processing communication signals is provided and may comprise, quantizing a channel estimate at a MIMO receiver onto a codebook based on a cost function, wherein the codebook comprises a plurality of unitary matrices. A codebook index may be fed back from the MIMO receiver to a MIMO transmitter, wherein the codebook index is associated with one of the plurality of unitary matrices that minimizes the cost function. The cost function may be minimized by choosing a smallest scalar cost from a plurality of scalar costs, wherein each one of the scalar costs is generated from one or more sums and one or more products of elements of a product matrix, and wherein the product matrix is generated from one of the plurality of unitary matrices, a matrix comprising the channel estimate, and their respective Hermitian Transposes.
Owner:AVAGO TECH INT SALES PTE LTD
Channel Quantization for Multiuser Diversity
ActiveUS20110206153A1Spatial transmit diversityError prevention/detection by diversity receptionMultiuser systemSystem capacity
Owner:AVAGO TECH INT SALES PTE LTD
Effective capacity optimization method of multi-antenna transmission under feedback limitation
ActiveCN108667498AIncrease effective capacityRadio transmissionChannel state informationWireless transmission
The invention discloses an effective capacity optimization method of multi-antenna transmission under feedback limitation. Firstly, a source node pre-codes the information, and then performs data transmission with a destination node in a beam forming manner. In the precoding of the source node, channel state information needs to be obtained, so that the destination node performs channel estimationat first, and feeds back the estimated channel state information to the source node after quantifying the same. As the accuracy of channel quantization is higher, the precoding performance of the source node is better, but this also increases the feedback cost. By application of the effective capacity optimization method of multi-antenna transmission under feedback limitation disclosed by the invention, the optimal channel feedback quantization bit is sought, and the effectiveness and the reliability of wireless transmission are improved under the condition of limited feedback cost, and the effective capacity of the multi-antenna transmission is significantly improved at the same time.
Owner:NANJING UNIV OF POSTS & TELECOMM
Method and system for an efficient channel quantization method for MIMO pre-coding systems
InactiveUS7822102B2Electric signal transmission systemsEqual length code transmitterChannel state informationUnitary matrix
Owner:AVAGO TECH INT SALES PTE LTD
Time domain modulation method based on channel encryption
ActiveCN110798832AImprove securityImprove confidentialityKey distribution for secure communicationChannel estimationTime domainCommunications system
The invention discloses a time domain modulation method based on channel encryption, which combines channel encryption and time domain modulation to realize a physical layer security scheme, and relates to a physical layer security algorithm in distributed wireless communication. According to the scheme, a distributed communication system communicates with a target node through distributed nodes,channel information of the distributed nodes and the target node is estimated and quantized, then a receiving party and a sending party generate secret keys according to the channel quantization information, and encryption is conducted by using the secret keys as time domain modulation parameters to achieve physical layer safety communication. The scheme can be applied to a distributed communication system, the safety of wireless communication is improved from a physical layer, and the interception probability of enemies to our information is effectively reduced.
Owner:NO 54 INST OF CHINA ELECTRONICS SCI & TECH GRP
Method and device for reducing complexity of channel quantization
ActiveUS20190068252A1Reduce complexityTransmission path divisionSignal allocationChannel state informationTime domain
Embodiments of the present invention provide a method and device for reducing complexity of channel quantization. The method comprises: determining, on the basis of whether a measurement feedback parameter configured by a channel state information (CSI) process satisfies a predetermined condition, a CSI report time-domain position and a time interval requirement parameter Nrpt-ref for a CSI reference resource time-domain position, or determining a CSI trigger time-domain position and a time interval requirement parameter Nrpt-ref for a CSI reference resource time-domain position; determining, according to the determined time interval requirement parameter Nrpt-ref and the CSI report time-domain position, the CSI reference resource time-domain position, or determining, according to the determined time interval requirement parameter Nrpt-ref and the CSI trigger time-domain position, the CSI reference resource time-domain position; and performing CSI quantization according to the determined CSI reference resource time-domain position.
Owner:ZTE CORP
Deep convolutional neural network quantization method, system and device, and storage medium
PendingCN114418062AImproving Quantitative Inference AccuracyImprove computing efficiencyNeural architecturesNeural learning methodsAlgorithmRandomized rounding
The invention provides a deep convolutional neural network quantization method, system and device, and a storage medium, and the method comprises the steps: analyzing a deep learning neural network model, and inserting a pseudo-quantization operator at an operator needing to be quantized; floating point type parameter reasoning is carried out on the constructed test set, and a feature quantization scaling factor is calculated according to the relative entropy divergence; quantizing weight parameters of a convolutional layer or a full connection layer of the deep learning neural network model to obtain a weight quantization scaling factor; and performing quantization reasoning according to the feature quantization scaling factor and the weight quantization scaling factor. According to the method, random rounding is applied to deep neural network model quantization, meanwhile, a channel quantization algorithm is optimized, quantization reasoning is carried out on designed special equipment, and the quantization reasoning precision and efficiency of the deep neural network model are improved.
Owner:山东云海国创云计算装备产业创新中心有限公司
Channel quantization for multiuser diversity
InactiveUS7936808B2Spatial transmit diversityError preventionChannel state informationSystem capacity
Owner:AVAGO TECH WIRELESS IP SINGAPORE PTE
Channel quantization information acquisition method, terminal, base station, and system
The invention provides a channel quantization information acquisition method, a terminal, a base station, and a system. The method comprises the following steps: according to a Channel State IndicatorReference Signal (CSI-RS) sent by a sending end, acquiring sequences of channels between a receiving end and the sending end, and acquiring codon matrixes according to the sequences and codon spaces;acquiring weight amplitude information corresponding to the codon matrixes, wherein the weight amplitude information comprises diagonal matrixes, the number of which is identical to the number of thesequences, and is used for amplitude adjustments on rows in the codon matrixes; and feeding the weight amplitude information and indication information corresponding to codon matrixes back to the sending end, so that the sending end can acquire a precoding matrix corresponding to the channels between the receiving end and the sending end according to the indication information, the weight amplitude information and a first matrix, and use the precoding matrix as channel quantization information. The method, the terminal, the base station and the system provided by the invention has the advantages that module values of weight values in the precoding matrix are no more identical due to the introduction of the weight amplitude information, so that the system performance can be improved, and the advantages of the dimension reduction feedback technology can be fully exerted.
Owner:BEIJING UNIV OF POSTS & TELECOMM
Quantizing a joint-channel-encoded audio signal
Provided are, among other things, systems, methods and techniques for quantizing a joint-channel-encoded audio signal, e.g., by: identifying a target quantization unit for reduction of quantization step size based on quantization errors; determining whether the target quantization unit has been jointly sum / difference encoded with another quantization unit; if the target quantization unit has been jointly sum / difference encoded with another quantization unit, then (i) designating the sum or difference channel quantization unit as a target S / D quantization unit in based on which has a greater quantization error and (ii) re-quantizing the target S / D channel quantization using a decreased quantization step size; recalculating the quantization error for the target quantization unit; and repeating the process until a specified criterion is satisfied.
Owner:DIGITAL RISE TECH CO LTD
Power distribution method and device
ActiveCN109981152ASolve the problem of application scopeRadio transmissionWireless communicationDistribution methodTerminal equipment
The invention discloses a power distribution method and equipment, which are used for solving the problem that the application range of an RS transmission strategy is limited. The method comprises thefollowing steps of: a network device obtains K channel quantization error mean values, wherein the K channel quantization error mean values are in one-to-one correspondence with the K terminal devices, and the ith channel quantization error mean value in the K channel quantization error mean values is determined according to an estimation channel vector and a quantization channel vector corresponding to a signal transmitted between the ith terminal device and the network device; and the network device determines a power distribution parameter value, wherein the power distribution parameter value is determined according to the K channel quantization error mean values. By adopting the method, the RS transmission strategy does not depend on an RVQ codebook, and the method is suitable for more codebooks.
Owner:HUAWEI TECH CO LTD
Quantization processing method and device and quantization processing chip
The embodiment of the invention provides a quantization processing method and device, and a quantization processing chip, and the method comprises the steps: firstly, determining the initial quantization parameter of each channel based on the distribution characteristics of the network parameters of each channel, and enabling the initial quantization parameter of one channel to be the optimal quantization parameter theoretically meeting the distribution characteristics of the network parameters of the channel; however, the initial quantization parameter may not satisfy a layer quantization hardware deployment condition; therefore, the optimized quantization parameter corresponding to each channel is searched in the search space determined on the basis of the initial quantization parameter, the quantization performance of the searched optimized quantization parameter is close to that of the initial quantization parameter, and a layer quantization hardware deployment condition can be met, so that the method can be applied to a general quantization processing chip, and the quantization performance of the universal quantization processing chip is close to that of a channel quantization mode.
Owner:BEIJING SENSETIME TECH DEV CO LTD
CVBS signal compression/decompression method, CVBS signal transmitting/receiving device and CVBS signal transmitting/receiving system
ActiveCN105812841AReduce occupancyLittle change in informationTelevision system detailsColor television detailsLeading edgeSignal compression
An embodiment of the invention discloses a CVBS signal compression / decompression method, a CVBS signal transmitting / receiving device and a CVBS signal transmitting / receiving system that are applied to the technical field of video signal processing. A large quantity of signals can be transmitted with a certain bandwidth, and a bandwidth utilization rate can be improved. The CVBS signal compression method mainly comprises the following steps: a first step of detecting quantified CVBS signals, a second step of determining leading edge data of a current synchronization pulse, and a third step of compressing data contained in the CVBS signals in order. The third step comprises the following steps: preset code words used for indicating the synchronization pulse are written into a first FIFO storage device, and a channel quantization value and a color load wave quantization value are directly written into the first FIFO storage device; compression values, determined via a compression table, corresponding to all quantization values between the color load wave and a next synchronization pulse are written into the first FIFO storage device; the compression table is used for storing mapping relations between the quantization values and compression values.
Owner:CAPITAL MICROELECTRONICS
Deep learning model reasoning method, system and equipment and computer medium
PendingCN114444658AImprove inference accuracyNeural architecturesInference methodsAlgorithmTheoretical computer science
The invention discloses a deep learning model reasoning method, system and device, and a computer medium. The method comprises the following steps: determining channel quantization parameters of each channel in a target deep learning model; determining a branch quantization parameter of the model branch based on the channel quantization parameter according to a relationship between a channel in the target deep learning model and the model branch; quantizing the target deep learning model based on the branch quantization parameters to obtain a quantized deep learning model; and reasoning based on the quantized deep learning model to obtain a reasoning result. In the application, the channel quantization parameter of each channel in the target deep learning module is determined, fine-grained quantization parameter determination is realized, the branch quantization parameter of the model branch is determined based on the channel quantization parameter, reasoning is performed based on the quantized deep learning model, and the reasoning result is obtained. The deep learning model reasoning based on the fine-grained channel quantization parameters is realized, and the reasoning precision of the deep learning model is improved.
Owner:INSPUR SUZHOU INTELLIGENT TECH CO LTD
Characteristic extraction method, system and device based on sub-channel quantization and storage medium
ActiveCN111614358ASolve technical problems that are difficult to adaptMitigate current performance degradationCode conversionCharacter and pattern recognitionVisual technologyFeature extraction
The embodiment of the invention relates to the technical field of computer vision, and discloses a feature extraction method, system and device based on sub-channel quantization and a storage medium.The method comprises the steps: firstly obtaining a to-be-processed image; inputting the to-be-processed image into the target feature extraction model to obtain target object features output by the target feature extraction model, wherein the target feature extraction model is a compressed feature extraction model obtained by performing model compression on floating point model parameters in theinitial feature extraction model in a mode of determining quantized decimal digits through different channels. Therefore, according to the embodiment of the invention, the initial feature extraction model is subjected to model compression operation, so the compressed model can adapt to the low-power-consumption equipment, and the technical problem that the low-power-consumption equipment is difficult to adapt to the floating point model is solved. Meanwhile, the quantized decimal digits are processed in a sub-channel mode, quantization losses can be reduced, and therefore the current situationthat the network performance is reduced can be relieved.
Owner:合肥的卢深视科技有限公司
Relay cooperative communication system information transmission method under feedback limitation
ActiveCN108768470AIncrease effective capacitySite diversityChannel state informationCommunications system
The invention discloses a relay cooperative communication system information transmission method under feedback limitation. The method comprises the steps that a source node sends information to relaynodes, and the relay nodes decode received information, pre-codes decoded information according to channel state information fed back by a destination node, maximizes a received signal to noise ratioof the destination node, and calculates precoding channel transmission capacity; the destination node carries out channel estimation to obtain channel state information, quantizes the channel state information, feeds back the information to the relay nodes and calculates feedback channel transmission capacity; and the difference between the precoding channel transmission capacity and the feedbackchannel transmission capacity is defined as effective capacity, the optimum feedback channel quantization bit is sought with the purpose of maximizing the effective capacity, and the pre-coded information is forwarded to the destination node. According to the method, through seeking of the optimum feedback channel quantization bit, the effective capacity of wireless relay transmission is remarkably improved.
Owner:NANJING UNIV OF POSTS & TELECOMM
A power distribution method and device
ActiveCN109981152BSolve the problem of application scopeRadio transmissionWireless communicationDistribution methodTerminal equipment
The invention discloses a power distribution method and equipment, which are used for solving the problem that the application range of an RS transmission strategy is limited. The method comprises thefollowing steps of: a network device obtains K channel quantization error mean values, wherein the K channel quantization error mean values are in one-to-one correspondence with the K terminal devices, and the ith channel quantization error mean value in the K channel quantization error mean values is determined according to an estimation channel vector and a quantization channel vector corresponding to a signal transmitted between the ith terminal device and the network device; and the network device determines a power distribution parameter value, wherein the power distribution parameter value is determined according to the K channel quantization error mean values. By adopting the method, the RS transmission strategy does not depend on an RVQ codebook, and the method is suitable for more codebooks.
Owner:HUAWEI TECH CO LTD
An efficient capacity optimization method for multi-antenna transmission under feedback constraints
ActiveCN108667498BIncrease effective capacityRadio transmissionChannel state informationWireless transmission
Owner:NANJING UNIV OF POSTS & TELECOMM
Method and device for channel state information feedback and receiving
ActiveCN102263619BHigh quantitative accuracyImprove performanceError prevention/detection by using return channelError prevention/detection by diversity receptionChannel state information feedbackControl theory
Owner:DATANG MOBILE COMM EQUIP CO LTD
Method for quantizing weight by channels
PendingCN113762499AIncrease profitEliminate the effects ofNeural architecturesNeural learning methodsAlgorithmNeural network nn
The invention provides a method for quantizing weight by channels, which is characterized in that the weight is quantized according to the number of output channels of a model, when the weight of a convolutional neural network is four-dimensional (height, width, input_channel and output_channel), the extreme values of other three-dimensional data are respectively counted according to the output channel, then the data are quantized into low bits, and the data are quantized according to the distribution characteristic of each channel. The invention aims to overcome the defects in the prior art and solve the problem that the precision of the model is reduced due to the fact that low-bit data is not fully utilized and concentrated in a small number of numerical values when the existing low-bit model is quantized.
Owner:合肥君正科技有限公司
Transmission beamforming method and apparatus in an orthogonal frequency division multiplexing-based MIMO wireless system
ActiveUS8873664B2Reduce errorsReduce overheadSpatial transmit diversityModulated-carrier systemsCode bookEngineering
Owner:SEOUL NAT UNIV R&DB FOUND
Channel randomization circuit and method based on time-interleaved ADC
PendingCN114244362AAvoid early outputImprove spurious-free dynamic rangeAnalogue-digital convertersDistribution controlControl signal
The invention provides a channel randomization circuit and method based on a time-interleaved ADC, and the circuit comprises a channel selection module which is used for outputting M clock receiving control signals and N coded data receiving control signals according to a main clock and a generated random number; the multi-phase clock distribution module is used for generating N multi-phase clocks according to a sampling main clock, redistributing the multi-phase clocks according to the clock receiving control signal, and outputting M redistributed clock signals; the time-interleaved ADC module is used for outputting M pieces of output data and a corresponding number of channel quantization completion signals according to the redistribution clock signal; the adjustable delay module is used for setting the delay size of the data receiving control signal; and the time sequence distribution control module is used for controlling the output data to be sequentially arranged and output according to a time sequence according to the delayed data receiving control signal and the channel quantization completion signal.
Owner:CHONGQING GIGACHIP TECH CO LTD
Method and Network for Time-Frequency Channel Attention Weight Calculation and Vectorization
This application relates to the method and network of time-frequency channel attention weight calculation and vectorization, including: using spherical embedding normalization to preprocess the logarithmic Fourier magnitude spectrum; segmenting the logarithmic Fourier magnitude spectrum preprocessing Process data into short-term spectral segments to generate multi-channel tensors; use the attention mechanism to calculate the attention weight of each channel in the time direction and frequency direction; obtain time-frequency attention by matrix multiplication of the attention weights in the time direction and frequency direction Weight coefficient tensor; use the attention mechanism to calculate the attention weight tensor of each channel in the time-frequency attention weight coefficient tensor; use the attention weight tensor of each channel to obtain vectorization in the time and frequency directions through one-dimensional convolution Results; concatenate the vectorized results of the time direction and frequency direction of at least one channel, and use one-dimensional convolution to obtain the channel vectorized results; calculate the average value of the channel vectorized results of the short-term spectrum segment and use it as the entire long-term The results of the time-to-speech correspondence.
Owner:INST OF AUTOMATION CHINESE ACAD OF SCI
Who we serve
- R&D Engineer
- R&D Manager
- IP Professional
Why Patsnap Eureka
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com