Fusion method and device of multi-source probability index data, equipment and medium
By calculating the risk of abandonment and generating weights, combined with Bayesian estimation, the problem of low accuracy and reliability of multi-source probability index data evaluation data is solved, and more accurate and reliable evaluation results are achieved.
Patent Information
- Application Number
- CN202411046023.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-11-26
AI Technical Summary
The prior art is difficult to effectively evaluate the internal and external field consistency of multi-source probability index data, resulting in low evaluation accuracy and reliability.
By obtaining the statistical values and probability of probability index data of multiple data sources, calculating the risk of truth abandonment, generating statistical credibility weights and experimental credibility weights, and calculating weights of weighted Bayesian estimation to ensure the evaluation results Reflects data consistency and user credibility.
The evaluation accuracy and reliability of multi-source probability index data is improved, and an evaluation result that comprehensively considers the importance and reliability of different data sources is provided.
Smart Images

Figure CN119026073A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of information processing technology, and in particular to a method, device, equipment and medium for fusing multi-source probability indicator data. Background Art
[0002] At present, for tests whose test results usually only include two results, success or failure, each test result corresponds to probability indicator data. Regarding the evaluation of such probability indicators, there is a lack of solutions that can verify the consistency of internal and external field data. In practical applications, multi-source probability indicator data are usually collected from different sources, such as probability indicator data collected from different sensors, tests, simulations, or actual battlefield environments. These data may have different accuracy, reliability, and observation errors, which leads to the introduction of bias and risk in the evaluation, and the accuracy and reliability of the evaluation are low. Summary of the invention
[0003] The present invention provides a method, device and medium for fusing multi-source probability indicator data, which can improve the evaluation accuracy and reliability of multi-source probability indicator data.
[0004] The technical solution of the present disclosure is achieved as follows:
[0005] In a first aspect, the present disclosure provides a method for fusing multi-source probability indicator data, the method comprising:
[0006] Obtaining a false positive risk according to statistical values and probabilities corresponding to multiple groups of probability indicator data obtained from multiple data sources, and obtaining a statistical credibility weight according to the false positive risk;
[0007] Determine the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data;
[0008] Obtaining a comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight;
[0009] Obtain parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values;
[0010] Obtaining the statistical weight corresponding to each group of probability indicator data according to the comprehensive weight of each group of probability indicator data and the Bayesian estimation intermediate quantity;
[0011] The weighted Bayesian estimate is obtained based on the statistical weights and parameter statistics corresponding to all probability indicator data.
[0012] In a second aspect, the present disclosure provides a device for fusing multi-source probability indicator data, the device comprising: a first acquisition part, a first determination part, a second acquisition part, a third acquisition part, a fourth acquisition part and a fifth acquisition part, wherein:
[0013] The first acquisition part is configured to acquire the risk of false positive rejection according to the statistical values and probabilities corresponding to the multiple groups of probability indicator data acquired from the multiple data sources, and acquire the statistical credibility weight according to the risk of false positive rejection;
[0014] The first determining part is configured to determine the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data;
[0015] The second acquisition part is configured to acquire the comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight;
[0016] The third acquisition part is configured to acquire parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values;
[0017] The fourth acquisition part is configured to acquire the statistical weight corresponding to each group of probability indicator data according to the comprehensive weight of each group of probability indicator data and the Bayesian estimation intermediate value;
[0018] The fifth acquisition part is configured to acquire a weighted Bayesian estimation value according to the statistical weights and parameter statistics corresponding to all probability-type indicator data.
[0019] In a third aspect, the present disclosure provides a computing device, comprising: a processor and a memory; the processor is used to execute instructions stored in the memory to implement the method for fusing multi-source probability-based indicator data as described in the first aspect.
[0020] In a fourth aspect, the present disclosure provides a computer-readable storage medium storing at least one instruction, wherein the at least one instruction is used to be executed by a processor to implement the method for fusing multi-source probability-based indicator data as described in the first aspect.
[0021] The present disclosure provides a method, device and medium for fusing multi-source probability indicator data. After obtaining the corresponding false negative risk α through the statistical values and probabilities of multiple groups of probability indicator data obtained from multiple data sources, the statistical credibility weight and the experimental credibility weight are generated in combination with the false negative risk α and the professional experience of the personnel, so as to quantify the reliability of each data source. Then, using relevant statistical means, the corresponding parameter statistics and the Bayesian estimated intermediate quantity are calculated according to the statistical value of each group of data. Then, the statistical weight is obtained by using the Bayesian estimated intermediate quantity and the comprehensive weight generated by the statistical credibility weight and the experimental credibility weight. The statistical weight takes into account the data consistency and user credibility, and ensures that the final weighted Bayesian estimate reflects the importance and reliability of different data sources, so that the technical solution provided by the present disclosure improves the evaluation accuracy and reliability of multi-source probability indicator data. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 A flowchart of a method for fusing multi-source probability indicator data provided by the present invention.
[0023] Figure 2 A schematic diagram of the process of obtaining the statistical credibility weight provided in the present disclosure.
[0024] Figure 3 A schematic diagram of the composition of a device for fusing multi-source probability indicator data provided by the present invention.
[0025] Figure 4 A schematic diagram of the hardware structure of a computing device provided by the present invention. DETAILED DESCRIPTION
[0026] The technical solutions in the present disclosure will be described clearly and completely below in conjunction with the accompanying drawings in the present disclosure.
[0027] Taking missiles or jammers as an example, their several indicators can be collected through simulation, actual battlefield, field test and other test methods. The test results usually only include two results: success or failure. In some examples, these two results can be 0 and 1 respectively. Each test result corresponds to probability indicator data, which can be used to quantify the key parameters of various performance and effects.
[0028] For example, in a missile interception test, only the test results of whether the missile successfully intercepts and destroys the target in a single test are included. In a missile single-shot kill test, only the test results of whether a single missile can hit and destroy the target in a single test are included. In a survival test, only the test results of whether the missile or jammer can be successfully launched and reach the target in a single test under enemy firepower are included. In a jamming test, only the test results of whether the jammer successfully jams the enemy radar in a single test are included. In an anti-jamming test, only the test results of whether the missile or jammer maintains normal operation in a single test under the condition of enemy electronic jamming are included. In a battlefield adaptability test, only the test results of whether the missile or jammer system can operate normally in a single test under different battlefield environments and conditions are included. The above-mentioned examples of tests correspond to probability indicator data, which are crucial for evaluating the effectiveness of combat systems, making tactical decisions, optimizing system design, and training and preparation. By collecting and analyzing these data, the reliability and effectiveness of missile or jammer systems can be improved, thereby gaining advantages in actual combat.
[0029] The above-mentioned probability indicator data usually come from different data sources. Each group of data sources provides a group of experimental data or test data. The value of each experimental data or test data in each group is 0 or 1. In this disclosure, the probability indicator data from different data sources are referred to as multi-source probability indicator data. In some examples, the data sources may include: different sensors, different experimental conditions or test scenarios, different observers, different geographical locations, different time points, etc. In order to integrate these probability indicator data from different data sources to provide a more accurate and comprehensive performance evaluation. See Figure 1 , which shows a method for fusing multi-source probability-based indicator data, the fusion method includes steps S110 to S160.
[0030] In step S110, a false positive risk is obtained according to statistical values and probabilities corresponding to multiple groups of probability indicator data obtained from multiple data sources, and a statistical credibility weight is obtained according to the false positive risk.
[0031] In the present disclosure, each data source provides a set of test result data as probability indicator data. For example, there are m data sources in total, and the test result data provided by the sth group of data sources can be expressed as Each data in this group is a single test result and can be 0 or 1. s Indicates the number of test result data provided by the s-th group of data sources. In the present disclosure, the test result data provided by the s-th group of data sources may also be referred to as the s-th group of probability indicator data. It can be understood that in the above example, m data sources will provide a total of m groups of probability indicator data.
[0032] In some examples, before executing S110, a consistency check can also be performed on multiple groups of probability indicator data provided by multiple data sources to ensure that the multiple groups of probability indicator data are statistically consistent, that is, they describe the same phenomenon or process and there is no significant statistical difference. For example, the probability indicator data collected by different data sources are independent of each other. The present disclosure can calculate the confidence interval corresponding to each group of probability indicator data respectively, and then determine whether there is a significant difference in the probability distribution of different groups of probability indicator data based on the chi-square test. In addition, it is also possible to compare whether different confidence intervals overlap. If there is a large overlap in the confidence intervals of all groups of probability indicator data, it can also be shown that they are statistically consistent. In addition, consistency checks can also be performed by other means, which will not be repeated in this disclosure. After multiple groups of probability indicator data have been subjected to consistency checks, it means that these data are statistically consistent and there is no significant difference, and step S110 can be continued according to these data. Otherwise, the inconsistent probability indicator data can be eliminated, and the remaining probability indicator data that meet the consistency can be used to continue to perform step S110. In this way, the higher risk of false information caused by inconsistent data can be avoided, and the accuracy and reliability of data fusion can be improved.
[0033] Generally speaking, the false negative risk α refers to the risk of misjudging an event that is actually positive as negative. In the present disclosure, the false negative risk corresponding to each group of probability indicator data is obtained for the statistical value and probability of each group of probability indicator data, and the statistical credibility weight corresponding to each group of probability indicator data is obtained by normalizing the false negative risk. Setting weights based on statistical analysis results can give higher weights to more reliable data, which helps to improve the evaluation accuracy of the fused data. Specifically, the false negative risk of the sth group of probability indicator data is α s , then the statistical credibility weight corresponding to this group
[0034] In step S120, the test credibility weight corresponding to each group of probability indicator data is determined by setting the test credibility for each group of probability indicator data.
[0035] In the present disclosure, in addition to the statistical analysis involved in step S110, the professional judgment and experience of the evaluators in actual scenarios are also taken into account. In some examples, the experimental credibility of each set of probability indicator data is set by the data evaluators or experimental testers, thereby introducing the patent knowledge and experience of these personnel, further improving the accuracy of the evaluation.
[0036] In some examples, data evaluators or experimental testers can set the experimental credibility for each of the m groups of probability indicator data, expressed as {c1, c2, ..., c m}, where c s is the experimental reliability of the sth group of data, and c s The value range of is 0 to 1. After the setting is completed, the test credibility c corresponding to the sth group of probability index data can be calculated. s Normalize according to the following formula to obtain the test credibility weight w corresponding to the sth group of probability index data cs :
[0037]
[0038] In step S130, the comprehensive weight of each group of probability indicator data is obtained according to the statistical credibility weight and the experimental credibility weight.
[0039] In the present disclosure, in combination with the aforementioned steps S110 and S120, for the sth group of probability indicator data, the corresponding statistical credibility weight is represented as w rs , and its corresponding test credibility weight is expressed as w cs , then the comprehensive weight w of the sth group of probability indicator data s It can be obtained by the following formula:
[0040]
[0041] Among them, m represents the number of groups of all probability indicator data.
[0042] It should be noted that in the present disclosure, the comprehensive weight not only takes into account the reliability of the data obtained by statistical analysis, but also takes into account the professional judgment of the user, providing a basis for obtaining more accurate weighted Bayesian estimates in subsequent steps.
[0043] In step S140, parameter statistics and Bayesian estimation intermediate quantities are obtained according to multiple groups of probability indicator data and their statistical values.
[0044] In the present disclosure, on the one hand, parameter statistics are used to provide input parameters for the weighted Bayesian estimation described in the final step S160. In some examples, the input parameters can use statistical related means to calculate the corresponding parameter statistics according to the statistical values of each group of data to ensure the accuracy of the subsequent weighted Bayesian estimation.
[0045] On the other hand, the Bayesian estimation intermediate quantity is used to calculate the subsequent statistical weights, and the statistical values corresponding to the multiple groups of probability indicator data and the sample size of the multiple groups of probability indicator data are used to obtain statistics that can reflect the combined influence of prior knowledge and observed data.
[0046] In step S150, the statistical weight corresponding to each group of probability indicator data is obtained according to the comprehensive weight of each group of probability indicator data and the Bayesian estimation intermediate value.
[0047] In the present disclosure, the weight of each set of data in the final estimate is calculated by using the comprehensive weights of the Bayesian estimation intermediate quantities and the basis, taking into account data consistency and user credibility, thereby ensuring that the final weighted Bayesian estimate reflects the importance and reliability of different data sources.
[0048] Specifically, for the sth group of probability indicator data, the corresponding comprehensive weight is w s , the corresponding Bayesian estimated intermediate quantity is T s , then the corresponding statistical weight is:
[0049]
[0050] Among them, i represents the i-th group of probability indicator data.
[0051] In step S160, a weighted Bayesian estimation value is obtained according to the statistical weights and parameter statistics corresponding to all probability indicator data.
[0052] In this disclosure, the statistics of each group of data are multiplied by the corresponding weights and then summarized to obtain the weighted Bayesian estimate. This generates a probability indicator evaluation result that fully considers data consistency, reliability, and user judgment. Figure 1 The scheme shown combines the principles of statistics and probability theory to provide a comprehensive indicator, namely the weighted Bayesian estimate. This indicator can help decision makers make more scientific and reasonable decisions when facing uncertainty and complexity, and improve the quality and effectiveness of decision-making.
[0053] pass Figure 1The technical solution shown in the invention obtains the corresponding false positive risk α through the statistical values and probabilities of multiple groups of probability indicator data obtained from multiple data sources, and then generates statistical credibility weights and test credibility weights in combination with the false positive risk α and the professional experience of the personnel, so as to quantify the reliability of each data source. Then, the corresponding parameter statistics and Bayesian estimated intermediate quantities are calculated according to the statistical values of each group of data using relevant statistical means. Then, the statistical weight is obtained using the Bayesian estimated intermediate quantity and the comprehensive weight generated by the statistical credibility weight and the test credibility weight. The statistical weight takes into account the data consistency and user credibility, thereby ensuring that the final weighted Bayesian estimate reflects the importance and reliability of different data sources. The invention calculates the statistical confidence through the sample size and binomial distribution, and combines it with the test credibility, and uses the Bayesian method for fusion, which greatly improves the rationality of the fusion result.
[0054] for Figure 1 The technical solution shown in the figure may be implemented in some possible ways, such as Figure 2 As shown, the step S110 described in which the risk of rejecting a true false positive is obtained according to the statistical values and probabilities corresponding to multiple groups of probability indicator data obtained from multiple data sources, and the statistical credibility weight is obtained according to the risk of rejecting a true false positive, includes steps S1101 to S1106.
[0055] In step S1101, statistical values and probabilities corresponding to multiple groups of probability-based indicator data are obtained.
[0056] Combination Figure 1 In the example described above, among the m groups of probability indicator data obtained from m data sources, the sth group of probability indicator data is For the sth group of probability indicator data, the present disclosure uses data and as the statistical value k corresponding to the sth group of probability indicator data s ,Right now And since the value of the probability indicator data involved in the present disclosure is 0 or 1, the probability corresponding to the sth group of probability indicator data is
[0057] In step S1102, the probability upper bound and the probability lower bound corresponding to each group of probability indicator data are obtained according to the probability corresponding to each group of probability indicator data and the expected error.
[0058] In the present disclosure, the probability p corresponding to the sth group of probability indicator data is obtained. s After that, the probability upper bound p corresponding to each set of probability indicator data can be calculated based on the expected deviation δ according to the following formula: s1 And the probability lower bound p s2 :
[0059] p s1 =p s-δ
[0060] p s2 =p s +δ
[0061] It can be understood that for each set of probability indicator data, the corresponding probability upper bound and probability lower bound can be obtained by the above method.
[0062] It should be noted that the probability upper bound and the probability lower bound provide an interval that can be used to quantify the uncertainty in the estimate of the true probability value, and when the interval lengths of the probability upper and lower bounds corresponding to a set of probability indicator data are wider, it indicates that the reliability of the data source of the set of probability indicator data is low, thereby introducing the reliability of the data into the statistical credibility weight.
[0063] In step S1103, according to the number of samples in each group of probability indicator data and the corresponding upper probability bound and lower probability bound, the number of hits corresponding to the upper probability bound and the lower probability bound of each group of probability indicator data are obtained.
[0064] In the present disclosure, taking the sth group of probability indicator data as an example, the number of hits corresponding to the probability upper bound and the probability lower bound can be calculated according to the following formula:
[0065] c s1 =floor(n s ×p s1 )
[0066] c s2 =ceil(n s ×p s2 )
[0067] Among them, c s1 represents the number of hits corresponding to the probability upper bound of the sth group of probability indicator data, c s2 It indicates the number of hits corresponding to the probability lower bound of the sth group of probability indicator data; floor indicates the rounding down operator, and ceil indicates the rounding up operator.
[0068] In step S1104, the area proportion outside the probability upper bound and probability lower bound corresponding to each group of probability indicator data is determined according to the number of hits and the probability corresponding to the probability upper bound and probability lower bound of each group of probability indicator data.
[0069] In the present disclosure, the area proportion α of the probability upper bound of the sth group of probability indicator data s1 And the proportion of areas outside the probability lower bound α s2 They can be calculated according to the following formulas:
[0070]
[0071] It should be noted that the area proportion α of the probability upper bound s1 And the proportion of areas outside the probability lower bound α s2 It can also evaluate the confidence level and quantify the uncertainty and potential risk. The larger the proportion, the more uncertainty and risk there is, and the lower the confidence level, which helps to understand the credibility of the data.
[0072] In step S1105, the false negative risk corresponding to each set of probability indicator data is obtained according to the proportion of the area outside the probability upper bound and the probability lower bound corresponding to each set of probability indicator data.
[0073] In the present disclosure, the area proportion α of the probability upper bound of the sth group of probability class indicator data is obtained s1 And the proportion of areas outside the probability lower bound α s2 Afterwards, we can follow α s =α s1 +α s2 Obtain the true rejection risk corresponding to the sth group of probability indicator data.
[0074] In step S1106, the rejection risk corresponding to each group of probability indicator data is normalized to obtain the statistical credibility weight corresponding to each group of probability indicator data.
[0075] It should be noted that the introduction of the risk of rejection into the calculation of statistical credibility weights quantifies the credibility and uncertainty of the data from a statistical perspective, thereby setting lower weights for probability indicator data with lower credibility or higher uncertainty, in order to improve the accuracy and reliability of subsequent evaluations.
[0076] for Figure 1 In some possible implementations of the technical solution shown, the step S140 of obtaining parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values includes:
[0077] Determine the parameter statistics corresponding to each group of probability indicator data according to the multiple groups of probability indicator data and the sample numbers of the multiple groups of probability indicator data;
[0078] The Bayesian estimated intermediate quantity corresponding to each group of probability indicator data is determined according to the statistical values corresponding to the multiple groups of probability indicator data and the sample quantity of the multiple groups of probability indicator data.
[0079] For the above implementation, in some examples, determining the parameter statistics corresponding to each group of probability indicator data according to the multiple groups of probability indicator data and the sample numbers of the multiple groups of probability indicator data includes:
[0080] For the sth group of probability indicator data, the parameter statistics corresponding to the sth group of probability indicator data are obtained according to the following formula
[0081]
[0082] Among them, m represents the number of groups of probability indicator data, n j Represents the sample size of the jth group of probability indicator data.
[0083] For the above implementation, in some examples, determining the Bayesian estimated intermediate quantity corresponding to each group of probability indicator data according to the statistical values corresponding to the multiple groups of probability indicator data and the number of samples of the multiple groups of probability indicator data includes:
[0084] For the sth group of probability indicator data, the Bayesian estimated intermediate quantity corresponding to the sth group of probability indicator data is obtained according to the following formula:
[0085]
[0086] where B(p,q) represents the Beta distribution, and k i Indicates the statistical value corresponding to the i-th group of probability indicator data, n s Represents the sample size of the sth group of probability indicator data.
[0087] Based on the above technical solution, the present disclosure uses three groups of probability indicator data from three data sources as examples to illustrate the fusion method of the above multi-source probability indicator data. It should be pointed out that for the sth group, the typical multi-source probability indicator data is expressed as Each data in this group is a single test result and can be 0 or 1. s represents the number of test result data provided by the sth group of data sources. s ,k s ), where Based on this representation, the three groups of probability indicator data are represented as (800, 600) for the first group, (200, 142) for the second group, and (20, 16) for the third group.
[0088] First, the statistical values and probabilities corresponding to the above three groups of probability indicator data are: k1 = 600, k2=144, k3=16,
[0089] Next, according to the probability corresponding to each group of probability indicator data and the expected deviation δ=0.05, the probabilities corresponding to the upper and lower bounds are calculated as follows:
[0090] p 11 =p1-δ=0.70, p 12 =p1+δ=0.80;
[0091] p 21 =p2-δ=0.67, p 22 =p2+δ=0.77;
[0092] p 31 =p3-δ=0.75, p 32 =p3+δ=0.85.
[0093] Then, according to the sample size of each group of probability indicator data and the probability values of the upper and lower limits, the number of hits corresponding to the upper and lower probability limits of each group of probability indicator data is calculated:
[0094] c 11 =floor(800×0.7)=560c 12 =ceil(800×0.8)=640;
[0095] c 21 =floor(200×0.67)=134c 22 =ceil(200×0.77)=154;
[0096] c 31 =floor(20×0.75)=15c 32 =ceil(20×0.85)=17.
[0097] Then, according to the binomial distribution formula, the proportion of the area outside the upper and lower bounds corresponding to each group of probability indicator data is calculated as follows:
[0098] α 11 =0.00077α 12 =0.00036;
[0099] α 21 =0.0680α 12 =0.0468;
[0100] α 21 =0.370α 12 =0.206.
[0101] Next, the rejection risk of each group of data is calculated based on the proportion of the area outside the upper and lower bounds corresponding to each group of probability indicator data:
[0102] α1=α11 +α 12 =0.00101;
[0103] α2=α 21 +α 22 =0.1048;
[0104] α3=α 31 +α 32 =0.576.
[0105] Then, according to the risk of false negatives, the normalized statistical credibility weights corresponding to each group of probability indicator data are calculated as follows: r1 =0.431,w r2 =0.386 and w r3 =0.183.
[0106] Then, according to the test credibility of each group of probability indicator data set by the user, the normalized test credibility weight is calculated. For example, if the credibility of three groups of probability indicator data is 0.7, 0.80, and 0.90 respectively, the corresponding test credibility weights are: w c1 =0.292, w c2 =0.333 and w c3 =0.375.
[0107] Next, according to the statistical credibility weight and the experimental credibility weight of each group of probability indicator data, the comprehensive weights of each group of probability indicator data are calculated as follows: w1=0.390, w2=0.398 and w3=0.212.
[0108] Then, the parameter statistics of each group of probability indicator data are calculated respectively. for: as well as
[0109] Then, the Bayesian estimated intermediate quantity T of each group of probability indicator data is calculated respectively. s for:
[0110] T1=3.711×10 -5 , T2=2.920×10 -5 and T3 = 3.227 × 10 -5 .
[0111] Then, according to the Bayesian estimation intermediate quantity and comprehensive weight calculation, the statistical weights corresponding to each group are: W1 = 0.439, W2 = 0.353 and W3 = 0.208
[0112] Finally, the weighted Bayesian estimate is obtained based on the statistical weights and parameter statistics corresponding to all probability indicator data.
[0113] It should be noted that Bayesian estimation can effectively integrate prior information and current test data, making the estimated post-validation value more reasonable.
[0114] Based on the same inventive concept as the above technical solution, see Figure 3 In addition to the multi-source probability index data fusion device 30 provided by the present disclosure, the device includes: a first acquisition part 301, a first determination part 302, a second acquisition part 303, a third acquisition part 304, a fourth acquisition part 305 and a fifth acquisition part 306, wherein,
[0115] The first acquisition part 301 is configured to acquire the risk of false negatives according to the statistical values and probabilities corresponding to the multiple groups of probability indicator data acquired from the multiple data sources, and acquire the statistical credibility weight according to the risk of false negatives;
[0116] The first determining part 302 is configured to determine the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data;
[0117] The second acquisition part 303 is configured to acquire the comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight;
[0118] The third acquisition part 304 is configured to acquire parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values;
[0119] The fourth acquisition part 305 is configured to acquire the statistical weight corresponding to each group of probability-based indicator data according to the comprehensive weight of each group of probability-based indicator data and the Bayesian estimation intermediate value;
[0120] The fifth acquisition part 306 is configured to acquire a weighted Bayesian estimation value according to the statistical weights and parameter statistics corresponding to all probability-type indicator data.
[0121] In some examples, the first acquisition part 301 is configured to:
[0122] Obtaining statistical values and probabilities corresponding to the multiple groups of probability indicator data;
[0123] Obtain the probability upper bound and probability lower bound corresponding to each group of probability indicator data according to the probability corresponding to each group of probability indicator data and the expected error;
[0124] According to the number of samples in each group of probability indicator data and the corresponding upper probability bound and lower probability bound, obtain the number of hits corresponding to the upper probability bound and the lower probability bound of each group of probability indicator data;
[0125] Determine the proportion of the area outside the probability upper bound and probability lower bound corresponding to each group of probability indicator data according to the number of hits and the probability corresponding to the probability upper bound and probability lower bound of each group of probability indicator data;
[0126] Obtain the rejection risk corresponding to each set of probability indicator data according to the proportion of areas outside the probability upper bound and probability lower bound corresponding to each set of probability indicator data;
[0127] The risk of false negatives corresponding to each group of probability indicator data is normalized to obtain the statistical credibility weight corresponding to each group of probability indicator data.
[0128] In some examples, the first determining portion 302 is configured to:
[0129] Setting a corresponding test credibility for each group of probability indicator data;
[0130] The test credibility corresponding to each group of probability indicator data is normalized to obtain the test credibility weight corresponding to each group of probability indicator data.
[0131] In some examples, the second obtaining part 303 is configured to:
[0132] By using the statistical credibility weight and the experimental credibility weight corresponding to each group of probability indicator data, the comprehensive weight of each group of probability indicator data is obtained by the following formula:
[0133]
[0134] Among them, w rs represents the statistical credibility weight corresponding to the sth group of probability indicator data, w cs It represents the experimental credibility weight corresponding to the sth group of probability indicator data, and m represents the number of groups of probability indicator data.
[0135] In some examples, the third acquisition part 304 is configured to:
[0136] Determine the parameter statistics corresponding to each group of probability indicator data according to the multiple groups of probability indicator data and the sample numbers of the multiple groups of probability indicator data;
[0137] The Bayesian estimated intermediate quantity corresponding to each group of probability indicator data is determined according to the statistical values corresponding to the multiple groups of probability indicator data and the sample quantity of the multiple groups of probability indicator data.
[0138] In some examples, the third acquisition part 304 is configured to:
[0139] For the sth group of probability indicator data, the parameter statistics corresponding to the sth group of probability indicator data are obtained according to the following formula
[0140]
[0141] Among them, m represents the number of groups of probability indicator data, n j Represents the sample size of the jth group of probability indicator data.
[0142] In some examples, the third acquisition portion 304 is configured to:
[0143] For the sth group of probability indicator data, the Bayesian estimated intermediate quantity corresponding to the sth group of probability indicator data is obtained according to the following formula:
[0144]
[0145] where B(p,q) represents the Beta distribution, and k i Indicates the statistical value corresponding to the i-th group of probability indicator data, n s Represents the sample size of the sth group of probability indicator data.
[0146] Please refer to Figure 4 , which shows a structural block diagram of a computing device provided by an exemplary embodiment of the present disclosure. In some examples, the computing device 40 can be at least one of a smart phone, a smart watch, a desktop computer, a laptop, a virtual reality terminal, an augmented reality terminal, a wireless terminal, and a laptop portable computer. The computing device 40 has a communication function and can access a wired network or a wireless network. The computing device 40 can generally refer to one of a plurality of terminals, and those skilled in the art can know that the number of the above terminals can be more or less. In some examples, the computing device 40 can receive data based on the wired network or wireless network to which it is connected. It can be understood that the computing device 40 undertakes the calculation and processing work of the technical solution of the present disclosure, and the present disclosure does not limit this.
[0147] like Figure 4 As shown, the computing device in the present disclosure may include one or more of the following components: a processor 410 and a memory 420 .
[0148] Optionally, the processor 410 uses various interfaces and lines to connect various parts of the entire computing device, and executes various functions of the computing device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 420, and calling data stored in the memory 420. Optionally, the processor 410 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 410 can integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), a neural network processor (NPU), and a baseband chip. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content that needs to be displayed on the touch display; the NPU is used to implement artificial intelligence (AI) functions; and the baseband chip is used to process wireless communications. It is understandable that the above baseband chip may not be integrated into the processor 410, but may be implemented by a separate chip.
[0149] The memory 420 may include a random access memory (RAM) or a read-only memory (ROM). Optionally, the memory 420 includes a non-transitory computer-readable storage medium. The memory 420 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 420 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above various method embodiments, etc.; the data storage area may store data created according to the use of the computing device, etc.
[0150] In addition, those skilled in the art will appreciate that the structure of the computing device shown in the above figures does not constitute a limitation on the computing device, and the computing device may include more or fewer components than shown, or combine certain components, or arrange the components differently. For example, the computing device also includes a display screen, a camera assembly, a microphone, a speaker, a radio frequency circuit, an input unit, a sensor (such as an acceleration sensor, an angular velocity sensor, a light sensor, etc.), an audio circuit, a WiFi module, a power supply, a Bluetooth module, and other components, which will not be described in detail here.
[0151] The present disclosure also provides a computer-readable storage medium storing at least one instruction, wherein the at least one instruction is used to be executed by a processor to implement the method for fusing multi-source probability-based indicator data as described in the above embodiments.
[0152] The present disclosure also provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium; a processor of a computing device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computing device executes to implement the method for fusing multi-source probability-based indicator data described in the above-mentioned embodiments.
[0153] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present disclosure can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any media that facilitates the transmission of computer programs from one place to another. Storage media can be any available media that can be accessed by a general or special-purpose computer.
[0154] It should be noted that the technical solutions described in the present disclosure can be combined arbitrarily without conflict.
[0155] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A fusion method for multi-source probability indicator data, It is characterized in that The fusion method comprises: Obtaining a false positive risk according to statistical values and probabilities corresponding to multiple groups of probability indicator data obtained from multiple data sources, and obtaining a statistical credibility weight according to the false positive risk; Determine the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data; Obtaining a comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight; Obtain parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values; Obtaining the statistical weight corresponding to each group of probability indicator data according to the comprehensive weight of each group of probability indicator data and the Bayesian estimation intermediate quantity; The weighted Bayesian estimate is obtained based on the statistical weights and parameter statistics corresponding to all probability indicator data.
2. The method according to claim 1, It is characterized in that The method of obtaining the risk of false positives based on the statistical values and probabilities corresponding to the multiple groups of probability indicator data obtained from multiple data sources, and obtaining the statistical credibility weight based on the false positives risk, includes: Obtaining statistical values and probabilities corresponding to the multiple groups of probability indicator data; Obtain the probability upper bound and probability lower bound corresponding to each group of probability indicator data according to the probability corresponding to each group of probability indicator data and the expected error; According to the number of samples in each group of probability indicator data and the corresponding upper probability bound and lower probability bound, obtain the number of hits corresponding to the upper probability bound and the lower probability bound of each group of probability indicator data; Determine the proportion of the area outside the probability upper bound and probability lower bound corresponding to each group of probability indicator data according to the number of hits and the probability corresponding to the probability upper bound and probability lower bound of each group of probability indicator data; Obtain the rejection risk corresponding to each set of probability indicator data according to the proportion of areas outside the probability upper bound and probability lower bound corresponding to each set of probability indicator data; The risk of false negatives corresponding to each group of probability indicator data is normalized to obtain the statistical credibility weight corresponding to each group of probability indicator data.
3. The method according to claim 1, It is characterized in that The step of determining the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data includes: Setting a corresponding test credibility for each group of probability indicator data; The test credibility corresponding to each group of probability indicator data is normalized to obtain the test credibility weight corresponding to each group of probability indicator data.
4. The method according to claim 1, It is characterized in that The step of obtaining the comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight includes: By using the statistical credibility weight and the experimental credibility weight corresponding to each group of probability indicator data, the comprehensive weight of each group of probability indicator data is obtained by the following formula: Among them, W rs represents the statistical credibility weight corresponding to the sth group of probability indicator data, w cs It represents the experimental credibility weight corresponding to the sth group of probability indicator data, and m represents the number of groups of probability indicator data.
5. The method according to claim 1, It is characterized in that The obtaining of parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values includes: Determine the parameter statistics corresponding to each group of probability indicator data according to the multiple groups of probability indicator data and the sample numbers of the multiple groups of probability indicator data; The Bayesian estimated intermediate quantity corresponding to each group of probability indicator data is determined according to the statistical values corresponding to the multiple groups of probability indicator data and the sample quantity of the multiple groups of probability indicator data.
6. The method according to claim 5, It is characterized in that The determining, according to the multiple groups of probability indicator data and the number of samples of the multiple groups of probability indicator data, the parameter statistics corresponding to each group of probability indicator data includes: For the sth group of probability indicator data, the parameter statistics corresponding to the sth group of probability indicator data are obtained according to the following formula Among them, m represents the number of groups of probability indicator data, n j Represents the sample size of the jth group of probability indicator data.
7. The method according to claim 5, It is characterized in that The determining, according to the statistical values corresponding to the multiple groups of probability indicator data and the sample quantity of the multiple groups of probability indicator data, the Bayesian estimated intermediate quantity corresponding to each group of probability indicator data comprises: For the sth group of probability indicator data, the Bayesian estimated intermediate quantity corresponding to the sth group of probability indicator data is obtained according to the following formula: where B(p,q) represents the Beta distribution, and k i Indicates the statistical value corresponding to the i-th group of probability indicator data, n s Represents the sample size of the sth group of probability indicator data.
8. A device for fusing multi-source probability indicator data. It is characterized in that The device comprises: a first acquisition part, a first determination part, a second acquisition part, a third acquisition part, a fourth acquisition part and a fifth acquisition part, wherein: The first acquisition part is configured to acquire the risk of false positive rejection according to the statistical values and probabilities corresponding to the multiple groups of probability indicator data acquired from the multiple data sources, and acquire the statistical credibility weight according to the risk of false positive rejection; The first determining part is configured to determine the test credibility weight corresponding to each group of probability indicator data by setting the test credibility for each group of probability indicator data; The second acquisition part is configured to acquire the comprehensive weight of each group of probability indicator data according to the statistical credibility weight and the experimental credibility weight; The third acquisition part is configured to acquire parameter statistics and Bayesian estimation intermediate quantities according to the multiple groups of probability indicator data and their statistical values; The fourth acquisition part is configured to acquire the statistical weight corresponding to each group of probability indicator data according to the comprehensive weight of each group of probability indicator data and the Bayesian estimation intermediate value; The fifth acquisition part is configured to acquire a weighted Bayesian estimation value according to the statistical weights and parameter statistics corresponding to all probability-type indicator data.
9. A computing device, It is characterized in that The computing device comprises: a processor and a memory; the processor is used to execute instructions stored in the memory to implement the method for fusing multi-source probability-based indicator data as described in any one of claims 1 to 7.
10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores at least one instruction, and the at least one instruction is used to be executed by a processor to implement the method for fusing multi-source probability-based indicator data as described in any one of claims 1 to 7.
