A data quality analysis method based on Bernoulli-Gaussian model and EM algorithm
Through the Bernoulli-Gaussian model and EM algorithm, a rough statistical model and a mixed model of geodetic measurement data are constructed, which solves the problem of threshold introduction in the existing technology, and realizes comprehensive analysis of geodetic measurement data and efficient outlier processing. Improve the scientificity and reliability of data quality analysis.
Patent Information
- Application Number
- CN202311354617.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-10-19
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2043-10-19
AI Technical Summary
The prior art requires the introduction of thresholds when dealing with outliers in geodesy measurements, resulting in the data quality analysis results that are not scientific and reliable enough and cannot avoid human intervention.
Using the data quality analysis method based on the Bernoulli-Gaussian model and EM algorithm, the mixed distribution of observations is calculated, the model parameters are estimated, the data quality is analyzed, and the threshold is used to distinguish outliers and normal values. .
A comprehensive analysis of geodetic data is realized, the accuracy information and rough error rate of observation values are provided, artificial intervention is avoided, and the scientificity and reliability of data quality analysis is improved.
Smart Images

Figure CN117473717B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of spatial geographic technology, in particular to a data quality analysis method based on a Bernoulli-Gaussian model and an EM algorithm. Background Art
[0002] In recent years, in modern geodetic practice, a large number of observations are recorded or sampled; it is almost impossible for such a data set to be free of outliers; therefore, outlier processing has become part of the daily work of geodesists; from a statistical point of view, it is believed that the occurrence of outliers is due to the inability of the observation model to provide an adequate fit or statistical explanation; Beckman and Cook (1983) divided the incompetence of the observation model into two categories: local model defects and global model defects; local model weaknesses are those that focus only on peripheral observations rather than the entire model; these reasons may require separate treatment of outliers because alternative models that include outliers are usually unknown; global model defects refer to reasons that cause the existing model to be replaced by a new or revised model of the entire sample; these reasons treat outliers as known properties and may lead to the replacement of the existing model with a hybrid model.
[0003] According to the causes of outliers, observation models can be divided into two categories; generally speaking, the causes of local model defects indicate that outliers should be treated independently from normal observations; in this case, in addition to the existing normal model to describe normal observation data, there will be an abnormal model to explain the existence of outliers; so far, there are many different normal-abnormal models, among which the two models proposed by Dixon (1950) are the most commonly used: one is called the mean shift model, which considers outliers to be the result of mean shift, and the other is called the variance inflation model, which regards outliers as the result of variance inflation (Lehmann et al. 2020); on the contrary, the causes of overall model defects require replacing the existing model with a new or revised model for the entire sample. In this case, a mixed model is often used to merge normal and abnormal observations; Hawkins (1980) can construct two commonly used mixed models, namely the location contamination model and the scale contamination model (Lehmann 2013); the former consists of a normal distribution and another location contaminated distribution, and the latter consists of a normal distribution and another scale contaminated distribution.
[0004] However, the current common solutions have many shortcomings, including: in geodesy, outliers are usually caused by gross errors, and gross errors usually cause outliers; therefore, it is natural to explain the causes of outliers in geodetic data processing by the nature of gross errors. In this case, the gross error model plays an important role in establishing the distinction and connection between different observation models. Summary of the invention
[0005] In view of the above problems that when processing outliers on data sets obtained in geodetic practice, the prior art needs to introduce thresholds to distinguish outliers from normal values and human intervention in data quality, the present invention is proposed.
[0006] Therefore, the problem to be solved by the present invention is how to provide a method for distinguishing abnormal values from normal values without introducing any threshold value, avoiding human intervention in data quality, and making the analysis results of data quality more scientific and reliable.
[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:
[0008] In the first aspect, an embodiment of the present invention provides a data quality analysis method based on a Bernoulli-Gaussian model and an EM algorithm, which includes preprocessing the collected geodetic data and constructing a Bernoulli-Gaussian model based on the statistics of gross errors; constructing a mixed model of observation values in a linear observation equation based on an Expectation Maximum algorithm; and calculating the linear observation equation based on the linear observation equation.
[0009] Bernoulli-Gaussian model parameters and analyze geodetic data quality.
[0010] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm described in the present invention, wherein: the gross error includes Bernoulli variables and Gaussian variables; the decomposition calculation of the gross error is as follows:
[0011]
[0012] Among them, e g is the gross error; Z is the pattern matrix; is a size vector.
[0013] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm of the present invention, wherein: the pattern matrix Z is a diagonal matrix with a value of 2 m The specific formula is as follows:
[0014] Z=Z i ,i∈{0,…,2 m -1}
[0015] where the j-th diagonal element of the pattern matrix Z follows the parameter ε j The Bernoulli distribution of the pattern matrix Z is as follows:
[0016]
[0017] Among them, Z ij Z i The j-th diagonal element of j is the gross error rate of the jth observation; the size vector It obeys multidimensional Gaussian distribution, and the specific formula is as follows:
[0018]
[0019] in, is the mean vector of gross error size; is the variance-covariance matrix.
[0020] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm of the present invention, the observed value includes the true value, the accidental error and the gross error, and the specific formula of the observed value is as follows:
[0021] y=Ax+e+e g
[0022] Where y is the observation vector; A is the non-random design matrix; x is the parameter vector to be estimated; and e is a random error vector with zero mean and variance Σ.
[0023] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm of the present invention, wherein: the construction of a mixed model of observation values in the linear observation equation comprises the following steps:
[0024] According to the additivity of Gaussian distribution, the probability distribution formula of the observed value y is as follows:
[0025]
[0026] in, is the probability distribution of the observed value y; x,Σ,ε, are the parameters to be estimated for the observation value y in the mixed model; the specific calculation steps for the parameters to be estimated for the observation value y in the mixed model are as follows: If the crude error rates of different observations in a type of observation equation are the same, the specific formula is as follows:
[0027] ε j =ε,j∈{1,…,m}
[0028] Will It is divided into two parts: known cofactor matrix and unknown factors. The specific formula is as follows:
[0029] Σ=σ 2 Q
[0030]
[0031]
[0032] Among them, Q, All are known cofactor vectors or matrices; the formula after the estimated parameters are transformed is as follows:
[0033]
[0034] in, is x,σ 2 ,ε, The probability distribution of y.
[0035] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm described in the present invention, wherein: the Bernoulli-Gaussian model parameters are calculated based on the linear observation equation using the Expectation Maximum algorithm, given x, σ 2 ,ε, The initial value of , and each iteration of the Expectation Maximum algorithm is divided into E step and M step.
[0036] As a preferred solution of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm of the present invention, the calculation formula of the E step is:
[0037]
[0038] Among them, γ(Z i ) is Z i The posterior probability of; the calculation formula of the M step is:
[0039]
[0040]
[0041] M i and N i The calculation formula is as follows:
[0042]
[0043] Substitute the parameter value on the right side of the M step into the estimation of the previous iteration, and use the parameter value on the left side of the M step as the new estimation of the parameter for this iteration until the iteration converges.
[0044] In the second aspect, in order to further solve the safety problems existing in geodetic measurement, the present invention provides a system for data quality analysis based on the Bernoulli-Gaussian model and the EM algorithm, which includes: a Bernoulli-Gaussian model module, which is used to decompose and calculate the gross error, and obtain the probability mass function and probability density function of whether the gross error occurs or not; a mixed model construction module, which is used to construct a mixed model of observation values in the linear observation equation; a parameter evaluation module, which is used to calculate the values of the Bernoulli-Gaussian model parameters in the linear observation equation and analyze the data quality of geodetic measurement.
[0045] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as described in the first aspect of the present invention is implemented.
[0046] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as described in the first aspect of the present invention.
[0047] The beneficial effects of the present invention are as follows: the present invention proposes a Bernoulli-Gaussian gross error statistical model, and based on the EM (Expectation Maximum) algorithm, proposes a method for estimating BG model parameters in a linear observation equation. The present invention can estimate BG model parameters in a single observation equation, and can also estimate BG model parameters in multiple observation equations; the present invention can not only obtain the accuracy information of the observation value, but also obtain the gross error rate, gross error size and other factors in the observation value, thereby providing a comprehensive analysis method for the data quality of geodetic measurement; the present invention does not need to introduce any threshold value for distinguishing abnormal values from normal values, thereby avoiding human intervention in data quality, and making the analysis result of data quality more scientific and reliable. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:
[0049] Figure 1 It is the PDF diagram estimated during the iteration process of the EM algorithm in Example 1.
[0050] Figure 2 Case 1 is the histogram and estimated PDF of the Gaussian model and the mixture model in different situations in Example 2.
[0051] Figure 3 The case 2 figure is the histogram and estimated PDF case 2 of the Gaussian model and the mixture model in different situations in Example 2.
[0052] Figure 4 Case 3 is the histogram and estimated PDF of the Gaussian model and the mixed model in different situations in Example 2. DETAILED DESCRIPTION
[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0055] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0056] Example 1
[0057] Reference Figure 1 , which is the first embodiment of the present invention, provides a data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm, comprising the following steps:
[0058] S1: Preprocess the collected geodetic data and construct a Bernoulli-Gaussian model based on the statistics of gross errors.
[0059] Preferably, the geodetic data is collected and preprocessed, and gross error statistics are performed.
[0060] Furthermore, the gross error includes Bernoulli variables and Gaussian variables; the decomposition of the gross error is calculated as follows: Wherein: the gross error includes Bernoulli variables and Gaussian variables; the decomposition of the gross error is calculated as follows:
[0061]
[0062] Among them, e g is the gross error; Z is the pattern matrix; is a size vector.
[0063] Furthermore, the pattern matrix Z is a diagonal matrix with a value of 2 m The specific formula is as follows:
[0064] Z=Z i ,i∈{0,…,2 m -1}
[0065] where the j-th diagonal element of the pattern matrix Z follows the parameter ε j The Bernoulli distribution of the pattern matrix Z is as follows:
[0066]
[0067] Among them, Z ij Z i The j-th diagonal element of j is the gross error rate of the jth observation.
[0068] Preferably, the size vector It obeys multidimensional Gaussian distribution, and the specific formula is as follows:
[0069]
[0070] in, is the mean vector of gross error size; is the variance-covariance matrix.
[0071] S2: Based on the Expectation Maximum algorithm, a mixed model of observations in the linear observation equation is constructed.
[0072] Preferably, the observed value includes the true value, accidental error and gross error, and the specific formula of the observed value is as follows:
[0073] y=Ax+e+e g
[0074] Where y is the observation vector; A is the non-random design matrix; x is the parameter vector to be estimated; and e is a random error vector with zero mean and variance Σ.
[0075] Furthermore, constructing a mixed model of observations in the linear observation equation includes the following steps: According to the additivity of Gaussian distribution, the probability distribution formula of the observation value y is as follows:
[0076]
[0077] in, is the probability distribution of the observed value y; x,Σ,ε, are the parameters to be estimated for the observed value y in the mixed model.
[0078] The specific calculation steps for the estimated parameters of the observation value y in the mixed model are as follows: If the crude error rates of different observation values in a type of observation equation are the same, the specific formula is as follows:
[0079] ε j =ε,j∈{1,…,m}
[0080] Σ, It is divided into two parts: known cofactor matrix and unknown factors. The specific formula is as follows:
[0081] Σ=σ 2 Q
[0082]
[0083]
[0084] Among them, Q, All are known cofactor vectors or matrices; the formula after the estimated parameters are transformed is as follows:
[0085]
[0086] in, is x,σ 2 ,ε, The probability distribution of y; specifically, for the linear observation equation y = Ax + e + e g , where A=[1,1,…,1] T ,x=[μ]; numerical examples are performed through Monte Carlo simulation (MCS); first, the true value of the observed value is simulated according to Ax, and random errors and gross errors are added to the true value of the observed value; the random errors of e are independent and completely obey the Gaussian distribution, that is, e~N(0,σ 2 );e gThe gross error in is simulated according to the BG model and The observations in y are samples drawn independently from the mixture distribution. Parameter estimation using the EM algorithm is a PDF fitting process for these samples.
[0087] S3: Calculate the Bernoulli-Gaussian model parameters based on the linear observation equation and analyze the data quality of geodetic measurements.
[0088] Preferably, the ExpectationMaximum algorithm is used to calculate the parameters of the Bernoulli-Gaussian model based on the linear observation equation. Given x,σ 2 ,ε, The initial value of , and each iteration of the Expectation Maximum algorithm is divided into E step and M step.
[0089] Furthermore, the calculation formula of step E is:
[0090]
[0091] Among them, γ(Z i ) is Z i The posterior probability of ; the calculation formula of the M step is:
[0092]
[0093] Among them, M i and N i The calculation formula is as follows:
[0094]
[0095] Substitute the parameter value on the right side of the M step into the estimation of the previous iteration, and use the parameter value on the left side of the M step as the new estimation of the parameter for this iteration until the iteration converges.
[0096] Specifically, by simulating n samples of the mixed model, the model parameters are calculated using the EM algorithm, and then the true values and initial values of different parameters in the mixed model are obtained as shown in Table 1:
[0097] Table 1. True values and initial values of different parameters in the hybrid model
[0098]
[0099] During the EM iteration process, the parameters calculated by the EM algorithm are shown in Table 2:
[0100] Table 2 Parameters estimated during the EM algorithm iteration process
[0101]
[0102] It can be seen from the results in Table 1 and Table 2 that the parameters calculated by the present invention are different in each iteration and finally converge to the true value.
[0103] Further, such as Figure 1 As shown in Figure 3, during the iterative process of EM, the calculated PDF gradually deforms and eventually approaches the true PDF.
[0104] This embodiment also provides a system for data quality analysis based on the Bernoulli-Gaussian model and the EM algorithm, including: a Bernoulli-Gaussian model module, used to decompose and calculate gross errors, and obtain a probability mass function and a probability density function of whether a gross error occurs; a mixed model construction module, used to construct a mixed model of observation values in a linear observation equation; a parameter evaluation module, used to calculate the values of Bernoulli-Gaussian model parameters in the linear observation equation, and analyze the data quality of geodetic measurement.
[0105] This embodiment also provides a computer device, which is suitable for the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm proposed in the above embodiment.
[0106] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0107] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.
[0108] In summary, the present invention proposes a Bernoulli-Gaussian gross error statistical model, and based on the EM (Expectation Maximum) algorithm, proposes a method for estimating the BG model parameters in a linear observation equation. The present invention can estimate the BG model parameters in a single observation equation, and can also estimate the BG model parameters in multiple observation equations; the present invention can not only obtain the accuracy information of the observation value, but also obtain the gross error rate, gross error size and other factors in the observation value, providing a comprehensive analysis method for the data quality of geodetic measurement; the present invention does not need to introduce any threshold value for distinguishing outliers from normal values, avoids human intervention in data quality, and makes the analysis results of data quality more scientific and reliable.
[0109] Example 2
[0110] Reference Figure 2 to Figure 4 , which is the second embodiment of the present invention. This embodiment is different from the first embodiment in that, in order to verify its beneficial effects, experimental comparison data between the present invention and the prior art are provided.
[0111] The following is a comparison between the present invention and a mixed model using three different parameter values and using a conventional Gaussian model and corresponding model parameters to calculate LS, as shown in Table 3, the true values and estimated values of LS and EM parameters under different conditions.
[0112] Table 3 True values and estimated values of LS and EM parameters under different conditions
[0113]
[0114]
[0115] From the results in Table 3, we can see that when the total error rate and the total error size are relatively small (see Case 1), there is no significant difference between the true value and the estimated parameters of LS and EM. As the total error rate increases and the total error size increases (see Case 2 and Case 3), the accuracy of LS estimation decreases significantly due to lack of robustness. On the contrary, the parameters estimated by EM are less affected by outliers and remain stable. When the proportion of outliers is large, the accuracy of gross error parameters is even improved.
[0116] like Figure 2 to Figure 4 As shown in the figure, when the total error rate and the total error size are relatively small, the PDFs estimated by the Gaussian model and the mixed model are close to the histogram, showing satisfactory performance in fitting these samples; with the increase of the total error rate and the increase of the total error size (see Case 2 and Case 3), the PDF estimated by the Gaussian model plus LS is significantly deformed compared with the histogram, and the Gaussian model loses the ability to fit samples contaminated by coarse errors of larger proportions and amplitudes; on the contrary, the PDF of the mixed model is always close to the histogram and maintains good fitting performance.
[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm, characterized in that: include: Preprocessing is performed based on the collected geodetic data, and a Bernoulli-Gaussian model is constructed based on the statistics of gross errors; Based on the Expectation Maximum algorithm, a mixed model of observations in the linear observation equation is constructed; Calculate Bernoulli-Gaussian model parameters based on linear observation equations and analyze geodetic data quality; The construction of a mixed model of observations in a linear observation equation comprises the following steps: According to the additivity of Gaussian distribution, the probability distribution formula of the observed value y is as follows: in, is the probability distribution of the observed value y; x,Σ,ε, are the parameters to be estimated for the observed value y in the mixed model; Z ij Z i The j-th diagonal element of j is the gross error rate of the jth observation; A is the non-random design matrix; The specific calculation steps for the estimated parameters of the observation value y in the mixed model are as follows: If the crude error rates of different observations in a type of observation equation are the same, the specific formula is as follows: e j =ε,j∈{1,…,m} Σ, It is divided into two parts: known cofactor matrix and unknown factors. The specific formula is as follows: S = s 2 Q Among them, Q, are all known cofactor vectors or matrices; The formula after the estimated parameters are transformed is as follows: in, is x,σ 2 ,ε, The probability distribution of y; The Bernoulli-Gaussian model parameters are calculated based on the linear observation equation using the Expectation Maximum algorithm. Given x, σ 2 ,ε, The initial value of , and each iteration of the Expectation Maximum algorithm is divided into E step and M step; The calculation formula of the E step is: Among them, γ(Z i ) is Z i The posterior probability of The calculation formula of the M step is: M i and N i The calculation formula is as follows: Substitute the parameter value on the right side of the M step into the estimation of the previous iteration, and use the parameter value on the left side of the M step as the new estimation of the parameter for this iteration until the iteration converges.
2. The data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as claimed in claim 1, characterized in that: The gross error includes Bernoulli variables and Gaussian variables; the decomposition of the gross error is calculated as follows: Among them, e g is the gross error; Z is the pattern matrix; is a size vector.
3. The data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as claimed in claim 2, characterized in that: The pattern matrix Z is a diagonal matrix with a value of 2 m The specific formula is as follows: Z=Z i ,i∈{0,…,2 m -1} where the j-th diagonal element of the pattern matrix Z follows the parameter ε j The Bernoulli distribution of the pattern matrix Z is as follows: Among them, Z ij Z i The j-th diagonal element of j is the gross error rate of the jth observation; The size vector It obeys multidimensional Gaussian distribution, and the specific formula is as follows: in, is the mean vector of gross error size; is the variance-covariance matrix.
4. The data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm as claimed in claim 3, characterized in that: The observed value includes the true value, accidental error and gross error. The specific formula of the observed value is as follows: y=Ax+e+e g Where y is the observation vector; A is the non-random design matrix; x is the parameter vector to be estimated; and e is a random error vector with zero mean and variance Σ.
5. A system for data quality analysis based on Bernoulli-Gaussian model and EM algorithm, based on the method for data quality analysis based on Bernoulli-Gaussian model and EM algorithm according to any one of claims 1 to 4, characterized in that: Also includes, Bernoulli-Gaussian model module, used to decompose and calculate gross errors and obtain the probability mass function and probability density function of whether gross errors occur or not; Mixed model building module, used to build mixed models of observations in linear observation equations; The parameter evaluation module is used to calculate the values of the Bernoulli-Gaussian model parameters in the linear observation equation and analyze the data quality of geodetic measurements.
6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm are implemented in any one of claims 1 to 4.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the data quality analysis method based on the Bernoulli-Gaussian model and the EM algorithm described in any one of claims 1 to 4 are implemented.
Citation Information
Patent Citations
Multi-gross error detection method and device based on maximum posteriori estimation and medium
CN114117802A
