WO2017118189A1 - 信号处理方法、信号处理装置及信号处理系统 - Google Patents

信号处理方法、信号处理装置及信号处理系统 Download PDF

Info

Publication number
WO2017118189A1
WO2017118189A1 PCT/CN2016/104111 CN2016104111W WO2017118189A1 WO 2017118189 A1 WO2017118189 A1 WO 2017118189A1 CN 2016104111 W CN2016104111 W CN 2016104111W WO 2017118189 A1 WO2017118189 A1 WO 2017118189A1
Authority
WO
WIPO (PCT)
Prior art keywords
signal
signal processing
global
nodes
sample
Prior art date
Application number
PCT/CN2016/104111
Other languages
English (en)
French (fr)
Inventor
全智
张洁
Original Assignee
南方科技大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 南方科技大学 filed Critical 南方科技大学
Publication of WO2017118189A1 publication Critical patent/WO2017118189A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/22Matching criteria, e.g. proximity measures
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12Classification; Matching

Definitions

  • the present invention relates to the field of signal processing, and in particular, to a signal processing method, a signal processing device, and a signal processing system.
  • Signal classification detection refers to the purpose of confirming the classification or mode of the input signal by extracting useful information in the received signal.
  • Signal classification detection is involved in many fields such as cognitive radio, sensor network, image processing, pattern recognition, speech signal recognition, fingerprint recognition, seismic signal analysis, radar signal detection and medical diagnosis.
  • cognitive radio sensor network
  • image processing pattern recognition
  • speech signal recognition fingerprint recognition
  • seismic signal analysis radar signal detection
  • radar signal detection medical diagnosis
  • the present invention aims to solve at least one of the technical problems existing in the prior art.
  • the main object of the present invention is to provide a signal processing method aimed at improving the accuracy of signal classification.
  • An embodiment of the present invention provides a signal processing method, where the signal processing method includes:
  • a determining step of determining the sample signal most similar to the input signal based on the global similarity value to determine a type of the input signal a determining step of determining the sample signal most similar to the input signal based on the global similarity value to determine a type of the input signal.
  • the global similarity value is obtained by processing the input signals of the global nodes to determine the type of the input signal, thereby improving the accuracy of the classification of the input signal.
  • the embodiment of the invention further provides a signal processing device for signal classification, the signal processing device comprising:
  • the receiving module comprising a plurality of nodes
  • control module configured to control the plurality of nodes to receive an input signal
  • the processing module includes a calculation module and a determination module, the calculation module is configured to compare the input signal with the sample signal to calculate a similarity value of each of the nodes, and to use according to each of the nodes
  • the similarity value calculates a global similarity value
  • the processing module is configured to calculate a plurality of the global similarity values corresponding to the sample signal one-to-one according to the plurality of sample signals;
  • the determining module is configured to determine the sample signal that is most similar to the input signal according to the global similarity value to determine a type of the input signal.
  • the embodiment of the present invention may calculate the similarity value of the node by using the above formula, or may be other processing or calculation manner to obtain the similarity value.
  • the computing module is further configured to provide global weights for each of the nodes and calculate the global similarity values based on the global weights.
  • the computing module can be further configured to calculate the global similarity value using the following formula:
  • i is a positive integer and is less than or equal to the number of types of the sample signal
  • k is a positive integer and is less than or equal to the number K of the nodes
  • G i is the global similarity value corresponding to the i-th sample signal.
  • ⁇ i,k is the global weight of the kth node, The similar value of the kth node.
  • the computing module can be further configured to calculate the similarity value for each of the nodes using the following formula:
  • i is a positive integer and is less than or equal to the number of types of the sample signal
  • k is a positive integer and is less than or equal to the number of nodes
  • s i (n) is the sample signal of the i-th type, x (k) ) (n) of the k-th node said input signal.
  • the signal processing device includes a prompting module for issuing a prompt.
  • the determining module is further configured to provide a similar threshold and generate a prompt signal when all of the global similar values are less than the similar threshold, and the control module is configured to control the prompt according to the prompt signal The module issues a prompt.
  • the determining module is further configured to provide a similarity threshold corresponding to the global similarity value, and to select the largest global similarity value among the global similarity values that are greater than or equal to the similarity threshold
  • the corresponding phase signal type is used as the type of the input signal.
  • the signal processing device includes a display module for displaying a type of the input signal.
  • Embodiments of the present invention also provide a signal processing system including the above signal processing apparatus.
  • FIG. 1 is a flow chart of a signal processing method according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of functional blocks of a signal processing device according to an embodiment of the present invention.
  • FIG. 3 is a flow chart of a signal processing method according to another embodiment of the present invention.
  • FIG. 4 is a flowchart of a signal processing method according to still another embodiment of the present invention.
  • FIG. 5 is a schematic diagram of functional blocks of a signal processing apparatus according to another embodiment of the present invention.
  • FIG. 6 is a flowchart of a signal processing method according to still another embodiment of the present invention.
  • FIG. 7 is a flowchart of a signal processing method according to another embodiment of the present invention.
  • FIG. 8 is a flowchart of a signal processing method according to still another embodiment of the present invention.
  • FIG. 9 is a schematic diagram of functional blocks of a signal processing system according to still another embodiment of the present invention.
  • FIG. 10 is a block diagram of a signal processing system in accordance with one embodiment of the present invention.
  • FIG. 11 is a statistical diagram of a signal processing method according to an embodiment of the present invention.
  • first and second are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated.
  • features defining “first” or “second” may include one or more of the described features either explicitly or implicitly.
  • the meaning of "a plurality" is two or more unless specifically defined otherwise.
  • the "on" or “below” of the second feature may include direct contact of the first and second features, and may also include the first sum, unless otherwise specifically defined and defined.
  • the second feature is not in direct contact but through additional features between them.
  • the first feature “above”, “above” and “above” the second feature includes the first feature directly above and above the second feature, or merely indicating that the first feature level is higher than the second feature.
  • the first feature “below”, “below” and “below” the second feature includes the first feature directly above and above the second feature, or merely the first feature level being less than the second feature.
  • Embodiments of the present invention provide a signal processing method for signal classification.
  • signal processing methods of embodiments of the present invention can be used to classify signals.
  • a signal processing method includes the following steps:
  • S10 providing multiple nodes and controlling multiple nodes to receive input signals
  • S30 Determine a sample signal most similar to the input signal according to the global similarity value to determine the type of the input signal.
  • the signal processing method of the embodiment of the present invention processes and determines a plurality of global nodes, and finally obtains the type of the input signal received by the node in the global.
  • the multi-node collaborative processing method is used to judge the type of the input signal, thus improving the accuracy of the signal classification.
  • an embodiment of the present invention further provides a signal processing apparatus 10 that can process signals to classify signals.
  • signal processing device 10 includes the following modules:
  • the receiving module 110 includes a plurality of nodes
  • control module 120 is configured to control multiple nodes to receive an input signal
  • generating module 130 configured to generate a plurality of sample signals
  • Step S10 can be implemented by the receiving module 110 and the control module 120.
  • Step S20 can be implemented by the generating module and the processing module, and step S30 can be implemented by the processing module.
  • the processing module 140 includes a calculation module 142 and a determination module 144.
  • the calculation module 142 can be used to compare the input signal with the sample signal to calculate the similarity value of each node, and can be used to calculate the global similarity according to the similarity value of each node.
  • the value calculation module 142 can be configured to calculate a plurality of the global similar values corresponding to the sample signals according to the plurality of sample signals, and the determining module 144 is configured to determine, according to the global similarity values, a sample that is most similar to the input signal.
  • the example signal determines the type of input signal.
  • the degree of similarity of the sample signal and the input signal needs to be quantified, and different quantization algorithms may be involved according to different needs.
  • quantification may be performed by computing similar values for nodes in a communication system to obtain a similarity of the input signal to the sample signal.
  • the amplitude and phase of the input signal can be extracted according to the amplitude-frequency characteristic and the phase-frequency characteristic of the input signal to be compared with the sample signal, thereby determining the similarity between the input signal and the sample signal.
  • the signal processing method of the embodiment of the present invention may further include the following steps:
  • S22 Provide global weights of each node and calculate global similarity values according to global weights.
  • the weights of the respective nodes of the receiving module 110 are different, so the global similarity value is calculated according to the global weight of the similarity value of each node.
  • step S22 can be implemented by a computing module 142 for providing global weights for each node and calculating global similar values based on global weights.
  • similar values of nodes can be calculated The global similarity value is then calculated.
  • the calculation module 142 can calculate global similar values using the following formula:
  • i is a positive integer and is less than or equal to the number of types of the sample signal
  • k is a positive integer and is less than or equal to the number K of the nodes
  • G i is a global similar value corresponding to the i-th sample signal
  • ⁇ i , k is the global weight of the kth node
  • the global weight ⁇ i,k of the kth node can be calculated.
  • it can be calculated by means of an algorithm. Of course, you can also get it with other calculation tools.
  • the correlation similarity threshold ⁇ i and the global weight ⁇ i,k occupied by each node in the system are calculated.
  • an optimization algorithm may be proposed to obtain the above two parameters, where ⁇ is a vector containing K elements and K is the number of nodes.
  • the processing module 140 determines the type of input signal is also the probability of H m:
  • the probability here is only to calculate the global weight ⁇ i,k and the similarity threshold ⁇ i . Obviously, in the embodiment of the present invention, after the global weight ⁇ i,k and the similar threshold ⁇ i are obtained, the global similarity value G i can be obtained.
  • the probability that the processing module 140 misjudges the input signal type as H j is:
  • the problem can be reduced to a condition limited by M-1 linear equations.
  • a non-convex problem of a quadratic equation condition It can be changed to a quadratic inequality condition by the relaxation quadratic equation condition, and this non-convex function can be transformed into a convex problem.
  • ⁇ i is the covariance of the noise.
  • T represents matrix transposition.
  • the Q -1 ( ⁇ ) function represents the inverse of the Gaussian white noise probability density function.
  • step S20 may employ a cross-correlation processing method to quantify the degree of similarity of the two signals.
  • the calculation module 142 can be used to calculate the similarity value of each node, and the similarity value of each node can be calculated by using the following formula:
  • i is a positive integer and is less than or equal to the number of types of the sample signal
  • k is a positive integer and is less than or equal to the number of nodes K
  • s i (n) is the i-th sample signal
  • x (k) (n) is The input signal of the kth node.
  • n ⁇ 0, 1, 2, ..., N-1 ⁇ , indicating the sampling point of the signal.
  • the similarity value of the node is obtained to calculate the global similarity value
  • the calculated object is the similarity value of the multiple nodes, so that the subsequent steps can globally process and judge, and the accuracy of the signal classification is improved.
  • the method of calculating the similarity value of each node is not limited to the above calculation formula.
  • the signal processing method may further include the following steps:
  • S40 Provide a similar threshold and issue a prompt when all global similar values are less than the similar threshold.
  • the signal processing device 10 can include a prompting module 160.
  • the prompt module 160 is used to issue a prompt.
  • step S40 can be implemented by the determining module 144, the control module 120, and the prompting module.
  • the determining module 144 is further configured to provide a similar threshold and generate a prompt signal when all the global similar values are less than the similar threshold
  • the control module 120 is configured to control the prompting module 160 to issue a prompt according to the prompt signal.
  • the determining step S30 may further include the following steps:
  • S301 Select a type of the sample signal corresponding to the largest global similarity value as the type of the input signal.
  • step S301 can be implemented by the determining module 144, and the determining module 144 is configured to select the type of the sample signal corresponding to the largest global similarity value as the type of the input signal.
  • the signal processing method also includes providing similar thresholds for comparison.
  • the steps of the signal processing method further include the following steps:
  • a similar threshold is provided and a type of the sample signal corresponding to the largest global similarity value is selected as a type of the input signal among global similarities greater than or equal to the similar threshold.
  • the similarity threshold is used to judge the degree of similarity between the input signal and the sample signal. By setting a reasonable similarity threshold, the sample signal corresponding to the global similarity value equal to or greater than the similar threshold can be considered to be similar to the input signal.
  • the signal, while the sample signal corresponding to the global similarity value less than the similar threshold belongs to a signal that is less similar to the input signal.
  • the most similar sample signals can be further selected in the similar signals and the signal type is determined as the type of the input signal, and if the global similarity values of all the sample signals are smaller than Similar thresholds, that is, all sample signals are not very similar to the input signal and can be considered unrecognizable and prompt.
  • Similar thresholds that is, all sample signals are not very similar to the input signal and can be considered unrecognizable and prompt.
  • the input signal may be subject to channel and noise during transmission and may be distorted. Therefore, each time the input signal type is judged, it is necessary to make a certain judgment on the measured global similarity value G i . Specifically, the following formula can be used to judge:
  • ⁇ i is a similar threshold.
  • this judgment formula may indicate that the input signal is severely distorted. At this time, each node may need to receive the input signal again.
  • step S50 may be implemented by the determining module 144, and the determining module 144 is configured to provide a similar threshold and select a type of the sample signal corresponding to the largest global similar value as the input among the global similar values greater than or equal to the similar threshold.
  • the type of signal An algorithm can be designed to calculate the similarity threshold ⁇ i .
  • the similarity threshold ⁇ i exists as an important estimation parameter like the global weight ⁇ i,k .
  • the calculation module 142 can be used to calculate the two parameters, or can be calculated by other computing devices.
  • Signal processing system 20 includes signal processing device 10.
  • the signal processing device 10 of an embodiment of the present invention further includes a display module 160.
  • the display module can be used to display the final judgement classification result, ie the type of input signal.
  • H i may represent a signal type corresponding to all sample signals, where i is a positive integer and is less than or equal to the number M of the sample signals.
  • Agent1, Agent2, ..., AgentK respectively, represent the input signals of each node,
  • ⁇ i,k is the global weight of the kth node.
  • FIG. 10 illustrates the signal processing method of the embodiment of the present invention in the form of a system block diagram. It is better shown that the embodiment of the present invention uses a synergistic effect of multiple nodes to process and classify signals.
  • FIG. 11 exemplifies the effect of the single-node signal processing classification and the correct judgment probability of using two nodes to perform signal processing classification.
  • the solid line in FIG. 11 represents a single node for signal processing classification method
  • the broken line represents a method for signal processing classification by two nodes, wherein P C is a probability for correctly judging signal classification, and P M is a probability of misclassification signal classification.
  • P C is a probability for correctly judging signal classification
  • P M is a probability of misclassification signal classification.
  • the figure illustrates that the method of using two nodes to coordinate signal processing classification is more accurate and reliable.
  • the signal processing method of the embodiment of the present invention uses the synergy of a plurality of nodes to determine the type of the input signal, which can effectively reduce the error rate and improve the achievability and stability of the system.
  • a "computer-readable medium” can be any apparatus that can contain, store, communicate, propagate, or transport a program for use in an instruction execution system, apparatus, or device, or in conjunction with the instruction execution system, apparatus, or device.
  • computer readable media include the following: electrical connections (electronic devices) having one or more wires, portable computer disk cartridges (magnetic devices), random access memory (RAM), Read only memory (ROM), erasable editable read only memory (EPROM or flash memory), fiber optic devices, and portable compact disk read only memory (CDROM).
  • the computer readable medium may even be a paper or other suitable medium on which the program can be printed, as it may be optically scanned, for example by paper or other medium, followed by editing, interpretation or, if appropriate, other suitable The method is processed to obtain the program electronically and then stored in computer memory.
  • portions of the embodiments of the invention may be implemented in hardware, software, firmware or a combination thereof.
  • multiple steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system.
  • a suitable instruction execution system For example, if implemented in hardware, as in another embodiment, Implemented by any one of the following techniques or combinations thereof; a discrete logic circuit having logic gates for implementing logic functions on data signals, an application specific integrated circuit having suitable combination logic gates, Programming Gate Array (PGA), Field Programmable Gate Array (FPGA), etc.
  • each functional unit in each embodiment of the present invention may be integrated into one processing module, or each unit may exist physically separately, or two or more units may be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or in the form of software functional modules.
  • the integrated modules, if implemented in the form of software functional modules and sold or used as stand-alone products, may also be stored in a computer readable storage medium.
  • the above mentioned storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

实施例公开了一种信号处理方法,用于信号分类,所述信号处理方法包括以下步骤:提供多个节点并控制所述多个节点接收输入信号(S10);将所述输入信号与样例信号比较以计算各所述节点的相似值,根据各所述节点的相似值计算全局相似值,提供多种所述样例信号以计算与所述样例信号一一对应的多个所述全局相似值(S20);根据所述全局相似值判断与所述输入信号最相似的所述样例信号以确定所述输入信号的类型(S30)。实施方式的信号处理方法对全局的多个节点进行处理及判断,最终得出全局中节点接收的输入信号的类型。采用多节点协同处理的方法对输入信号的类型进行判断,如此提高了信号分类的准确率。实施例还公开了一种信号处理装置(10)及信号处理系统(20)。

Description

信号处理方法、信号处理装置及信号处理系统
优先权信息
本申请请求2016年1月6日向中国国家知识产权局提交的、专利申请号为201610006242.0的专利申请的优先权和权益,并且通过参照将其全文并入此处。
技术领域
本发明涉及信号处理领域,特别涉及一种信号处理方法、信号处理装置及信号处理系统。
背景技术
信号分类检测是指通过提取接收信号中的有用信息,以达到确认输入信号的分类或模式的目的。在认知无线电、传感器网络、图像处理、模式识别、语音信号识别、指纹识别、地震信号分析、雷达信号检测和医疗诊断等许多领域都涉及到了信号分类检测。然而,按照常规算法进行信号分类时,会有较大概率出现误判。
发明内容
本发明旨在至少解决现有技术中存在的技术问题之一。
本发明的主要目的在于提供一种信号处理方法,旨在提高信号分类的准确率。
本发明实施方式提供一种信号处理方法,所述信号处理方法包括:
控制步骤,提供多个节点并控制所述多个节点接收输入信号;
比较步骤,将所述输入信号与样例信号比较以计算各所述节点的相似值,根据各所述节点的相似值计算全局相似值,提供多种所述样例信号以计算与所述样例信号一一对应的多个所述全局相似值;及
判断步骤,根据所述全局相似值判断与所述输入信号最相似的所述样例信号以确定所述输入信号的类型。
如此,通过对全局各个节点的输入信号的处理得到全局相似值以对所述输入信号的类型进行判断,提高了所述输入信号分类的准确率。
本发明实施方式还提供了一种信号处理装置,用于信号分类,所述信号处理装置包括:
接收模块,所述接收模块包括多个节点;
控制模块,所述控制模块用于控制所述多个节点接收输入信号;
生成模块,所述生成模块用于生成多种样例信号;及
处理模块,所述处理模块包括计算模块及判断模块,所述计算模块用于将所述输入信号与所述样例信号比较以计算各所述节点的相似值,及用于根据各所述节点的相似值计算全局相似值,所述处理模块用于根据所述多种样例信号计算与所述样例信号一一对应的多个所述全局相似值;
所述判断模块用于根据所述全局相似值判断与所述输入信号最相似的所述样例信号以确定所述输入信号的类型。
本发明实施方式可以采用上述公式计算所述节点的所述相似值也可以是其他的处理或者计算方式得到所述相似值。
在某些实施方式中,所述计算模块还用于提供各所述节点的全局权重并根据所述全局权重计算所述全局相似值。
在某些实施方式中,所述计算模块还可以用于采用以下公式计算所述全局相似值:
Figure PCTCN2016104111-appb-000001
其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数K,Gi为第i种样例信号对应的所述全局相似值,ωi,k为第k个所述节点的所述全局权重,
Figure PCTCN2016104111-appb-000002
为第k个所述节点的所述相似值。
在某些实施方式中,所述计算模块还可以用于采用以下公式计算各所述节点的所述相似值:
Figure PCTCN2016104111-appb-000003
其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数,si(n)为第i种所述样例信号,x(k)(n)为第k个所述节点的所述输入信号。
在某些实施方式中,所述信号处理装置包括提示模块,所述提示模块用于发出提示。
在某些实施方式中,所述判断模块还用于提供相似阈值并在所有所述全局相似值小于所述相似阈值时产生提示信号及所述控制模块用于根据所述提示信号控制所述提示模块发出提示。
在某些具体实施方式中,所述判断模块还用于提供与所述全局相似值对应的相似阈值并用于在大于等于所述相似阈值的所述全局相似值中选择最大的所述全局相似值对应的所述相例信号类型作为所述输入信号的类型。
在某些实施方式中,所述信号处理装置包括显示模块,所述显示模块用于显示所述输入信号的类型。
本发明实施方式还提供了一种信号处理系统,所述信号处理系统包括上述信号处理装置。
本发明的附加方面和优点将在下面的描述中部分给出,部分将从下面的描述中变得明显,或通过本发明的实践了解到。
附图说明
本发明的上述和/或附加的方面和优点从结合下面附图对实施方式的描述中将变得明显和容易理解,其中:
图1是本发明一个实施方式的信号处理方法的流程图;
图2是本发明一个实施方式的信号处理装置的功能模块示意图;
图3是本发明另一个实施方式的信号处理方法的流程图;
图4是本发明又一个实施方式的信号处理方法的流程图;
图5是本发明另一个实施方式的信号处理装置的功能模块示意图;
图6是本发明再一个实施方式的信号处理方法的流程图;
图7是本发明另一个实施方式的信号处理方法的流程图;
图8是本发明又一个实施方式的信号处理方法的流程图;
图9是本发明再一个实施方式的信号处理系统的功能模块示意图;
图10是本发明一个实施方式的信号处理系统的框图;及
图11是本发明一个实施方式的信号处理方法的统计图。
具体实施方式
下面详细描述本发明的实施方式,所述实施方式的示例在附图中示出,其中自始至终相同或类似的标号表示相同或类似的元件或具有相同或类似功能的元件。下面通过参考附图描述的实施方式是示例性的,仅用于解释本发明的实施方式,而不能理解为对本发明的实施方式的限制。
在本发明的实施方式的描述中,需要理解的是,术语“中心”、“纵向”、“横向”、“长度”、“宽度”、“厚度”、“上”、“下”、“前”、“后”、“左”、“右”、“竖直”、“水平”、“顶”、“底”、“内”、“外”、“顺时针”、“逆时针”等指示的方位或位置关系为基于附图所示的方位或位置关系,仅是为了便于描述本发明的实施方式和简化描述,而不是指示或暗示所指的装置或元件必须具有特定的方位、以特定的方位构造和操作,因此不能理解为对本发明的实施方式的限制。此外,术语“第一”、“第二”仅用于描述目的,而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此,限定有“第一”、“第二”的特征可以明示或者隐含地包括一个或者更多个所述特征。在本发明的实施方式的描述中,“多个”的含义是两个或两个以上,除非另有明确具体的限定。
在本发明的实施方式的描述中,需要说明的是,除非另有明确的规定和限定,术语 “安装”、“相连”、“连接”应做广义理解,例如,可以是固定连接,也可以是可拆卸连接,或一体地连接;可以是机械连接,也可以是电连接或可以相互通讯;可以是直接相连,也可以通过中间媒介间接相连,可以是两个元件内部的连通或两个元件的相互作用关系。对于本领域的普通技术人员而言,可以根据具体情况理解上述术语在本发明的实施方式中的具体含义。
在本发明的实施方式中,除非另有明确的规定和限定,第一特征在第二特征之“上”或之“下”可以包括第一和第二特征直接接触,也可以包括第一和第二特征不是直接接触而是通过它们之间的另外的特征接触。而且,第一特征在第二特征“之上”、“上方”和“上面”包括第一特征在第二特征正上方和斜上方,或仅仅表示第一特征水平高度高于第二特征。第一特征在第二特征“之下”、“下方”和“下面”包括第一特征在第二特征正上方和斜上方,或仅仅表示第一特征水平高度小于第二特征。
下文的公开提供了许多不同的实施方式或例子用来实现本发明的实施方式的不同结构。为了简化本发明的实施方式的公开,下文中对特定例子的部件和设置进行描述。当然,它们仅仅为示例,并且目的不在于限制本发明。此外,本发明的实施方式可以在不同例子中重复参考数字和/或参考字母,这种重复是为了简化和清楚的目的,其本身不指示所讨论各种实施方式和/或设置之间的关系。此外,本发明的实施方式提供了的各种特定的工艺和材料的例子,但是本领域普通技术人员可以意识到其他工艺的应用和/或其他材料的使用。
本发明实施方式提供了一种信号处理方法,用于信号分类。在某些实施方式中,本发明实施方式的信号处理方法可以用于对信号进行分类。
请参阅图1,本发明实施方式的信号处理方法包括以下步骤:
S10:提供多个节点并控制多个节点接收输入信号;
S20:将输入信号与样例信号比较以计算各节点的相似值,根据各节点的相似值计算全局相似值,提供多种样例信号以计算与样例信号一一对应的多个全局相似值;及
S30:根据全局相似值判断与输入信号最相似的样例信号以确定输入信号的类型。
本发明实施方式的信号处理方法对全局的多个节点进行处理及判断,最终得出全局中节点接收的输入信号的类型。采用多节点协同处理的方法对输入信号的类型进行判断,如此提高了信号分类的准确率。
请参阅图2,本发明实施方式还提供了一种信号处理装置10,信号处理装置10可以对信号进行处理以对信号进行分类。
在某些实施方式中,信号处理装置10包括以下模块:
接收模块110,接收模块110包括多个节点;
控制模块120,控制模块120用于控制多个节点接收输入信号;
生成模块130,生成模块130用于生成多种样例信号;及
处理模块140。
步骤S10可由接收模块110及控制模块120实现,步骤S20可由生成模块及处理模块实现,步骤S30可由处理模块实现。具体地,处理模块140包括计算模块142及判断模块144,计算模块142可以用于将输入信号与样例信号比较以计算各节点的相似值,及可以用于根据各节点的相似值计算全局相似值,计算模块142可以用于根据多种样例信号计算与样例信号一一对应的多个所述全局相似值,判断模块144用于根据所述全局相似值判断与输入信号最相似的样例信号以确定输入信号的类型。
在某些实施方式中,需要对样例信号和输入信号的相似程度进行量化,可以根据不同的需求涉及不同的量化算法。例如,在通信领域,进行量化可以是计算通信系统中个节点的相似值,从而得到输入信号与样例信号的相似程度。一般的,可以根据输入信号的幅频特性和相频特性提取该输入信号的幅值和相位以便与样例信号进行比较,从而判段输入信号与样例信号的相似度。
请参阅图3,本发明实施方式的信号处理方法还可包括以下步骤:
S22:提供各节点的全局权重并根据全局权重计算全局相似值。
在求取全局相似值时,接收模块110的各个节点的权重是不同的,因此要根据每个节点相似值的全局权重计算全局相似值。
在某些实施方式中,步骤S22可由计算模块142实现,计算模块142用于提供各节点的全局权重并根据全局权重计算全局相似值。
在某些实施方式中,可以在计算得到各节点相似值
Figure PCTCN2016104111-appb-000004
后再对全局相似值进行计算。
在某些实施方式中,计算模块142可以采用以下公式计算全局相似值:
Figure PCTCN2016104111-appb-000005
其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数K,Gi为第i种样例信号对应的全局相似值,ωi,k为第k个节点的全局权重,
Figure PCTCN2016104111-appb-000006
为第k个节点的相似值。
在某些实施方式中,第k个节点的全局权重ωi,k可以通过计算得到。例如,可以借助算法来计算得到。当然,还可以借助其他的计算工具得到。
在某些实施方式中,在样例信号分类为Hi时,计算得到相关相似阈值τi及各节点在系统中所占的全局权重ωi,k。在某些实施方式中可以提出优化算法以得到上述两个参数,其中ω为含有K个元素的向量,K为节点个数。
当输入信号分类实际为Hm时,处理模块140判断输入信号类型也为Hm的概率为:
Figure PCTCN2016104111-appb-000007
此处的概率只为了计算全局权重ωi,k和相似阈值τi。显然,在本发明实施方式中是有了全局权重ωi,k和相似阈值τi后,才可以求得全局相似值Gi
当信号分类实际为Hm时,处理模块140误判输入信号类型为Hj的概率为:
Figure PCTCN2016104111-appb-000008
最优化全局权重ωm和相似阈值τi使得处理模块140正确判断的概率最大,同时控制其误判概率在一定范围内,通常选取∈m,j<0.5。其中∈m,j是一常数值,出现在如下公式:
Figure PCTCN2016104111-appb-000009
以上公式是一个限制条件。
即:
Figure PCTCN2016104111-appb-000010
在某些实施方式中,考虑到系统的信道和高斯白噪声对输入信号的影响,结合高斯白噪声的概率密度函数特性,此问题可以归结为一个受限于M-1个线性等式条件和1个二次等式条件的非凸问题。可以通过松弛二次等式条件变为二次不等式条件,此非凸函数就可以转化为一个凸问题求解了。可以得到:
Figure PCTCN2016104111-appb-000011
这个方程可以使用许多软件的标准工具箱求解,如CVX。其中
Figure PCTCN2016104111-appb-000012
Figure PCTCN2016104111-appb-000013
Σi是噪声的协方差。符号T表示矩阵转置。Q-1(·)函数表示高斯白噪声概率密度函数的逆函数。
如此,就可以求解出最优的节点的全局权重ωm和相似阈值τm
Figure PCTCN2016104111-appb-000014
Figure PCTCN2016104111-appb-000015
在某些实施方式中,各节点的相似值的计算方式大有不同。例如,在通信领域中,步骤S20可以采用互相关的处理方法去量化两个信号的相似程度。
具体的,计算模块142可以用于计算各节点的相似值,各节点的相似值可以采用以下公式来计算得到:
Figure PCTCN2016104111-appb-000016
其中,i为正整数且小于等于样例信号的种类数,k为正整数且小于等于节点的个数K,si(n)为第i种样例信号,x(k)(n)为第k个节点的输入信号。此外,n={0,1,2,…N-1},表示信号的采样点。
如此,得到了节点的相似值才可对全局的相似值进行计算,且计算的对象是多个节点的相似值,以便后续步骤对全局进行处理及判断,提高了信号分类的准确率。此外,计算各节点的相似值的方法并不限制于上述计算公式。
请参阅图4,在本实施方式中,信号处理方法还可包括以下步骤:
S40:提供相似阈值并在所有全局相似值小于相似阈值时发出提示。
本发明实施方式的信号处理方法中,在所有全局相似值小于相似阈值时即可以判断样例信号中无与输入信号相似的信号。换言之,输入信号的类型与所有样例信号的类型不符。此时,可以发出提示以提示用户。
请参阅图5,信号处理装置10可包括提示模块160。提示模块160用于发出提示。
在某些具体实施方式中,步骤S40可由判断模块144、控制模块120及提示模块实现。具体地,判断模块144还用于提供相似阈值并在所有全局相似值小于相似阈值时产生提示信号及控制模块120用于根据提示信号控制提示模块160发出提示。
请参阅图6,本发明实施方式的信号处理方法中,判断步骤S30还可包括以下步骤:
S301:选择与最大的全局相似值对应的样例信号的类型为输入信号的类型。
本步骤采用上述实施方式的计算公式:
Figure PCTCN2016104111-appb-000017
Figure PCTCN2016104111-appb-000018
采用上述计算公式,全局相似值Gi越大,样例信号与输入信号的相似程度越大,因此,选择最大的全局相似值最大对应的样例信号的类型为输入信号的类型。
在某些实施方式中,步骤S301可采用判断模块144实现,判断模块144用于选择与最大的全局相似值对应的样例信号的类型为输入信号的类型。
为了能够更加准确的做出判断,信号处理方法还包括提供相似阈值进行比较。
请参阅图7,在某些实施方式中,信号处理方法的步骤还包括以下步骤:
S50:提供相似阈值并在大于等于相似阈值的全局相似值中选择最大的全局相似值对应的所述样例信号的类型作为所述输入信号的类型。
相似阈值的作用是对输入信号与样例信号之间的相似程度进行判断,通过设定合理的相似阈值,可认为大于等于相似阈值的全局相似值对应的样例信号属于与输入信号比较相似的信号,而小于相似阈值的全局相似值对应的样例信号属于与输入信号不太相似的信号。
请参阅图8,综合上述实施方式,可在比较相似的信号中进一步选择最相似的样例信号并将其信号类型认定为输入信号的类型,而若所有样例信号对应的全局相似值均小于相似阈值,也就是说,所有的样例信号均与输入信号不太相似,可认定为无法识别,并发出提示。这样设置的好处是,能保证选择出的与输入信号最相似的样例信号符合一定的相似条件,而当所有样例信号相似程度均不高时能作出无法识别的判定,从而满足用户的需求。
在某些实施方式中,输入信号在传输的过程中可能会受到信道和噪声的影响,从而会失真。因而在每次判断输入信号类型时,需要对衡量的全局相似值Gi进行一定判断。具体的,可以采用下面的公式进行判断:
Gi≥τi
其中,τi为相似阈值,当此判断公式不成立时,可能说明输入信号严重失真,此时,各节点可能需要重新接收输入信号。
在某些实施方式中,步骤S50可采用判断模块144实现,判断模块144用于提供相似阈值并在大于等于相似阈值的全局相似值中选择最大的全局相似值对应的样例信号的类型作为输入信号的类型。可以设计算法来计算相似阈值τi。在本发明实施方式的信号处理方法中,相似阈值τi与全局权重ωi,k一样作为重要的估计参数存在。可以采用计算模块142对着两个参数进行计算,也可以采用其他的计算设备进行计算。
如此,根据相似阈值进一步对全局相似值进行判断提高了对输入信号的类型判断的准确率,降低了信号分类的错误率。
请参阅图9,本发明还提供了一种信号处理系统20。信号处理系统20包括信号处理装置10。
本发明实施方式的信号处理装置10还包括显示模块160。在某些实施方式中,显示模块可以用于显示最终的判断分类结果,即输入信号的类型。
请参阅图10,图10中Hi可以表示所有样例信号对应的信号类型,其中i为正整数且小于等于样例信号的种类数M。Agent1,Agent2,……,AgentK,分别表示各个节点的输入信号,
Figure PCTCN2016104111-appb-000019
为第k个节点的相似值,ωi,k为第k个节点的全局权重。图10采用系统框图的形式对本发明实施方式的信号处理方法进行阐述,更好的表示了本发明实施方式是采用多个节点的协同作用对信号进行处理分类的。
请参阅图11,图11简单的例举了单节点进行信号处理分类与采用两个节点协同进 行信号处理分类的正确判断概率的影响。具体的,图11中的实线代表单节点进行信号处理分类方法,虚线代表两个节点协同进行信号处理分类方法,其中PC为正确判断信号分类的概率,PM为错误判断信号分类的概率。该图说明采用两个节点协同进行信号处理分类的方法的正确率更高更可靠。
本发明实施方式的信号处理方法采用多个节点的协同作用对输入信号的类型进行判断,可以有效的降低差错率,提高了系统的可实现性与稳定性。
在本说明书的描述中,参考术语“一个实施方式”、“一些实施方式”、“示意性实施方式”、“示例”、“具体示例”或“一些示例”等的描述意指结合所述实施方式或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施方式或示例中。在本说明书中,对上述术语的示意性表述不一定指的是相同的实施方式或示例。而且,描述的具体特征、结构、材料或者特点可以在任何的一个或多个实施方式或示例中以合适的方式结合。
流程图中或在此以其他方式描述的任何过程或方法描述可以被理解为,表示包括一个或更多个用于实现特定逻辑功能或过程的步骤的可执行指令的代码的模块、片段或部分,并且本发明的优选实施方式的范围包括另外的实现,其中可以不按所示出或讨论的顺序,包括根据所涉及的功能按基本同时的方式或按相反的顺序,来执行功能,这应被本发明的实施例所属技术领域的技术人员所理解。
在流程图中表示或在此以其他方式描述的逻辑和/或步骤,例如,可以被认为是用于实现逻辑功能的可执行指令的定序列表,可以具体实现在任何计算机可读介质中,以供指令执行系统、装置或设备(如基于计算机的系统、包括处理模块的系统或其他可以从指令执行系统、装置或设备取指令并执行指令的系统)使用,或结合这些指令执行系统、装置或设备而使用。就本说明书而言,"计算机可读介质"可以是任何可以包含、存储、通信、传播或传输程序以供指令执行系统、装置或设备或结合这些指令执行系统、装置或设备而使用的装置。计算机可读介质的更具体的示例(非穷尽性列表)包括以下:具有一个或多个布线的电连接部(电子装置),便携式计算机盘盒(磁装置),随机存取存储器(RAM),只读存储器(ROM),可擦除可编辑只读存储器(EPROM或闪速存储器),光纤装置,以及便携式光盘只读存储器(CDROM)。另外,计算机可读介质甚至可以是可在其上打印所述程序的纸或其他合适的介质,因为可以例如通过对纸或其他介质进行光学扫描,接着进行编辑、解译或必要时以其他合适方式进行处理来以电子方式获得所述程序,然后将其存储在计算机存储器中。
应当理解,本发明的实施方式的各部分可以用硬件、软件、固件或它们的组合来实现。在上述实施方式中,多个步骤或方法可以用存储在存储器中且由合适的指令执行系统执行的软件或固件来实现。例如,如果用硬件来实现,和在另一实施方式中一样,可 用本领域公知的下列技术中的任一项或他们的组合来实现:具有用于对数据信号实现逻辑功能的逻辑门电路的离散逻辑电路,具有合适的组合逻辑门电路的专用集成电路,可编程门阵列(PGA),现场可编程门阵列(FPGA)等。
本技术领域的普通技术人员可以理解实现上述实施例方法携带的全部或部分步骤是可以通过程序来指令相关的硬件完成,所述的程序可以存储于一种计算机可读存储介质中,该程序在执行时,包括方法实施例的步骤之一或其组合。
此外,在本发明的各个实施例中的各功能单元可以集成在一个处理模块中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个模块中。上述集成的模块既可以采用硬件的形式实现,也可以采用软件功能模块的形式实现。所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时,也可以存储在一个计算机可读取存储介质中。
上述提到的存储介质可以是只读存储器,磁盘或光盘等。
尽管上面已经示出和描述了本发明的实施例,可以理解的是,上述实施例是示例性的,不能理解为对本发明的限制,本领域的普通技术人员在本发明的范围内可以对上述实施例进行变化、修改、替换和变型。

Claims (16)

  1. 一种信号处理方法,用于信号分类,其特征在于,所述信号处理方法包括:
    控制步骤,提供多个节点并控制所述多个节点接收输入信号;
    比较步骤,将所述输入信号与样例信号比较以计算各所述节点的相似值,根据各所述节点的相似值计算全局相似值,提供多种所述样例信号以计算与所述样例信号一一对应的多个所述全局相似值;及
    判断步骤,根据所述全局相似值判断与所述输入信号最相似的所述样例信号以确定所述输入信号的类型。
  2. 如权利要求1所述的信号处理方法,其特征在于,所述比较步骤包括:
    计算子步骤,提供各所述节点的全局权重并根据所述全局权重计算所述全局相似值。
  3. 如权利要求2所述的信号处理方法,其特征在于,所述比较步骤采用以下公式计算所述全局相似值:
    Figure PCTCN2016104111-appb-100001
    其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数K,Gi为第i种样例信号对应的所述全局相似值,ωi,k为第k个所述节点的所述全局权重,
    Figure PCTCN2016104111-appb-100002
    为第k个所述节点的所述相似值。
  4. 如权利要求3所述的信号处理方法,其特征在于,所述比较步骤采用以下公式计算各所述节点的所述相似值:
    Figure PCTCN2016104111-appb-100003
    其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数,si(n)为第i种所述样例信号,x(k)(n)为第k个所述节点的所述输入信号。
  5. 如权利要求3所述的信号处理方法,其特征在于,所述信号处理方法包括:
    提示步骤,提供相似阈值并在所有所述全局相似值小于所述相似阈值时发出提示。
  6. 如权利要求3所述的信号处理方法,其特征在于,所述判断步骤包括选择与最大的所述全局相似值对应的所述样例信号的类型为所述输入信号的类型。
  7. 如权利要求3所述的信号处理方法,其特征在于,所述信号处理方法包括提供相似阈值并在大于等于所述相似阈值的所述全局相似值中选择最大的所述全局相似值对应的所述样例信号的类型作为所述输入信号的类型。
  8. 一种信号处理装置,用于信号分类,其特征在于,所述信号处理装置包括:
    接收模块,所述接收模块包括多个节点;
    控制模块,所述控制模块用于控制所述多个节点接收输入信号;
    生成模块,所述生成模块用于生成多种样例信号;及
    处理模块,所述处理模块包括计算模块及判断模块;
    所述计算模块用于将所述输入信号与所述样例信号比较以计算各所述节点的相似值,及用于根据各所述节点的相似值计算全局相似值,所述处理模块用于根据所述多种样例信号计算与所述样例信号一一对应的多个所述全局相似值;
    所述判断模块用于根据所述全局相似值判断与所述输入信号最相似的所述样例信号以确定所述输入信号的类型。
  9. 如权利要求8所述的信号处理装置,其特征在于,所述计算模块还用于提供各所述节点的全局权重并根据所述全局权重计算所述全局相似值。
  10. 如权利要求9所述的信号处理装置,其特征在于,所述计算模块采用以下公式计算所述全局相似值:
    Figure PCTCN2016104111-appb-100004
    其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数K,Gi为第i种样例信号对应的所述全局相似值,ωi,k为第k个所述节点的所述全局权重,
    Figure PCTCN2016104111-appb-100005
    为第k个所述节点的所述相似值。
  11. 如权利要求10所述的信号处理装置,其特征在于,所述计算模块还用于采用以下公式计算各所述节点的所述相似值:
    Figure PCTCN2016104111-appb-100006
    其中,i为正整数且小于等于所述样例信号的种类数,k为正整数且小于等于所述节点的个数,si(n)为第i种所述样例信号,x(k)(n)为第k个所述节点的所述输入信号。
  12. 如权利要求10所述的信号处理装置,其特征在于,所述信号处理装置块包括提示模块;
    所述判断模块还用于提供相似阈值并在所有所述全局相似值小于所述相似阈值时产生提示信号;及
    所述控制模块用于根据所述提示信号控制所述提示模块发出提示。
  13. 如权利要求10所述的信号处理装置,其特征在于,所述判断模块还用于选择与最大的所述全局相似值对应的所述样例信号类型为所述输入信号的类型。
  14. 如权利要求10所述的信号处理装置,其特征在于,所述判断模块还用于提供与所述全局相似值对应的相似阈值并用于在大于等于所述相似阈值的所述全局相似值中选择最大的所述全局相似值对应的所述相例信号类型作为所述输入信号的类型。
  15. 如权利要求8所述的信号处理装置,其特征在于,所述信号处理装置包括显示模块,所述显示模块用于显示所述输入信号的类型。
  16. 一种信号处理系统,用于信号分类,其特征在于,所述信号处理系统包括如权利要求8-15任意一项所述的信号处理装置。
PCT/CN2016/104111 2016-01-06 2016-10-31 信号处理方法、信号处理装置及信号处理系统 WO2017118189A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610006242.0A CN105678256B (zh) 2016-01-06 2016-01-06 信号处理方法、信号处理装置及信号处理系统
CN201610006242.0 2016-01-06

Publications (1)

Publication Number Publication Date
WO2017118189A1 true WO2017118189A1 (zh) 2017-07-13

Family

ID=56299080

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/104111 WO2017118189A1 (zh) 2016-01-06 2016-10-31 信号处理方法、信号处理装置及信号处理系统

Country Status (2)

Country Link
CN (1) CN105678256B (zh)
WO (1) WO2017118189A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105678256B (zh) * 2016-01-06 2019-01-18 南方科技大学 信号处理方法、信号处理装置及信号处理系统
CN111061163A (zh) * 2019-12-31 2020-04-24 北京黎明文仪家具有限公司 一种基于物联网的家用家具消防控制系统、方法及装置
CN113311258A (zh) * 2021-04-28 2021-08-27 深圳海关工业品检测技术中心 一种半电波暗室装置及电磁兼容emc测试方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103474065A (zh) * 2013-09-24 2013-12-25 贵阳世纪恒通科技有限公司 基于自动分类技术的语音意图判定识别方法
CN104347068A (zh) * 2013-08-08 2015-02-11 索尼公司 音频信号处理装置和方法以及监控系统
CN104952449A (zh) * 2015-01-09 2015-09-30 珠海高凌技术有限公司 环境噪声声源识别方法及装置
CN105678256A (zh) * 2016-01-06 2016-06-15 南方科技大学 信号处理方法、信号处理装置及信号处理系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104347068A (zh) * 2013-08-08 2015-02-11 索尼公司 音频信号处理装置和方法以及监控系统
CN103474065A (zh) * 2013-09-24 2013-12-25 贵阳世纪恒通科技有限公司 基于自动分类技术的语音意图判定识别方法
CN104952449A (zh) * 2015-01-09 2015-09-30 珠海高凌技术有限公司 环境噪声声源识别方法及装置
CN105678256A (zh) * 2016-01-06 2016-06-15 南方科技大学 信号处理方法、信号处理装置及信号处理系统

Also Published As

Publication number Publication date
CN105678256A (zh) 2016-06-15
CN105678256B (zh) 2019-01-18

Similar Documents

Publication Publication Date Title
US11715001B2 (en) Water quality prediction
EP3136297A1 (en) System and method for determining information and outliers from sensor data
WO2020155518A1 (zh) 物体检测方法、装置、计算机设备及存储介质
US10747637B2 (en) Detecting anomalous sensors
KR101734829B1 (ko) 지역성 말투를 구분하는 음성 데이터 인식 방법, 장치 및 서버
Granjon The CuSum algorithm-a small review
Yin et al. Optimal linear combinations of multiple diagnostic biomarkers based on Youden index
JP2023134499A (ja) ラベルノイズが存在する状態でのロバストな訓練
WO2017118189A1 (zh) 信号处理方法、信号处理装置及信号处理系统
US20170140273A1 (en) System and method for automatic selection of deep learning architecture
US20230120894A1 (en) Distance-based learning confidence model
WO2015085102A1 (en) System and method for non-invasive application recognition
US10108513B2 (en) Transferring failure samples using conditional models for machine condition monitoring
CN113657249B (zh) 训练方法、预测方法、装置、电子设备以及存储介质
WO2020173270A1 (zh) 用于分析数据的方法、设备和计算机存储介质
CN109960626A (zh) 端口异常的识别方法、装置、设备及介质
CN105187143B (zh) 一种基于二项分布的快速频谱感知方法和装置
US11907663B2 (en) Methods for unsupervised prediction of performance drop due to domain shift
CN108268467B (zh) 一种基于属性的异常数据检测方法和装置
US20150095490A1 (en) Online sparse regularized joint analysis for heterogeneous data
CN114067350A (zh) 模型更新方法、设备和计算机程序产品
US11080612B2 (en) Detecting anomalous sensors
CN112348161A (zh) 神经网络的训练方法、神经网络的训练装置和电子设备
CN114397964B (zh) 一种有效注视点检测方法、装置、电子设备及存储介质
US20230068381A1 (en) Method and electronic device for quantizing dnn model

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16883317

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16883317

Country of ref document: EP

Kind code of ref document: A1