Edge distribution acquisition method and device, equipment and storage medium thereof

Abstract

The application provides an edge distribution obtaining method, device and equipment and a storage medium thereof. The method comprises the following steps: obtaining an event occurrence quantity matrix corresponding to a conditional probability distribution of a target node of a Bayesian network; determining event occurrence quantities corresponding to a plurality of discrete values of the target node and a total event occurrence quantity of the target node according to the event occurrence quantity matrix corresponding to the conditional probability distribution; and obtaining an edge distribution of the target node by using a preset event quantity algorithm according to the total event occurrence quantity and the event occurrence quantities corresponding to the plurality of discrete values. Thus, the calculation amount of the nodes in the Bayesian network when calculating the edge probability is reduced, the calculation efficiency of the edge probability is improved, and technical support is provided for meeting the edge distribution calculation requirement in the massive data application scenario.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer application, and in particular, to an edge distribution obtaining method and device, a computer device, and a storage medium thereof. BACKGROUND

[0002] With the development of computer technology, a Bayesian network is defined by using two elements of network structure and conditional probability distribution, so as to meet the calculation requirement of the edge distribution of the Bayesian network in the application scenario of massive data by the Bayesian network. For example, the requirement of estimating the probability of a certain value under certain conditions in the data processing scenario, the requirement of estimating the possibility of a certain subject appearing in an image meeting a certain image feature in the image processing scenario, and the like.

[0003] However, after the structure learning and parameter learning of the discrete Bayesian network by using the joint probability distribution algorithm and the like, only the conditional probability distribution of each node is reserved as the parameter, and the original data is not reserved, so that when the current technology queries the probability of a certain node, the full nodes of the Bayesian network need to be searched and retrieved, the query efficiency is low, and when the network structure is large, the query efficiency is low or even unable to run, which cannot meet the edge distribution calculation requirement in the application scenario of massive data. SUMMARY

[0004] The present application aims to at least solve one of the technical problems in the related art to some extent.

[0005] To this end, a first object of the present application is to provide an edge distribution obtaining method to improve the calculation efficiency of the edge probability.

[0006] A second object of the present application is to provide an edge distribution obtaining device.

[0007] A third object of the present application is to provide a computer device.

[0008] A fourth object of the present application is to provide a non-transitory computer readable storage medium.

[0009] A fifth object of the present application is to provide a computer program product.

[0010] To achieve the above objects, an edge distribution obtaining method according to a first aspect of the present application comprises the following steps.

[0011] An event occurrence quantity matrix corresponding to the conditional probability distribution of the target node of the Bayesian network is obtained.

[0012] determine, according to the event number matrix corresponding to the conditional probability distribution, event numbers corresponding to a plurality of discrete values of the target node, and a total event number of the target node;

[0013] obtain, according to the total event number and the event numbers corresponding to the plurality of discrete values, an edge distribution of the target node by using a preset event number algorithm.

[0014] To achieve the above object, the second aspect of the present application provides an edge distribution obtaining device, comprising:

[0015] a first obtaining module, configured to obtain an event number matrix corresponding to a conditional probability distribution of a target node of a Bayesian network;

[0016] a determining module, configured to determine, according to the event number matrix corresponding to the conditional probability distribution, event numbers corresponding to a plurality of discrete values of the target node, and a total event number of the target node;

[0017] a second obtaining module, configured to obtain, according to the total event number and the event numbers corresponding to the plurality of discrete values, an edge distribution of the target node by using a preset event number algorithm.

[0018] To achieve the above object, the third aspect of the present application provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the edge distribution obtaining method of the first aspect of the present application is realized.

[0019] To achieve the above object, the fourth aspect of the present application provides a non-transitory computer readable storage medium, which stores a computer program, wherein when the processor executes the computer program, the edge distribution obtaining method of the first aspect of the present application is realized.

[0020] To achieve the above object, the fifth aspect of the present application provides a computer program product, wherein when the processor in the computer program product executes, the edge distribution obtaining method of the first aspect of the present application is realized.

[0021] The embodiments of the present application have at least the following additional technical effects:

[0022] The event occurrence quantity matrix corresponding to the conditional probability distribution of the target node of the Bayesian network is acquired, the event occurrence quantities corresponding to the plurality of discrete values of the target node and the total event occurrence quantity of the target node are determined according to the event occurrence quantity matrix corresponding to the conditional probability distribution, and then the marginal distribution of the target node is acquired by using a preset event quantity algorithm according to the total event occurrence quantity and the event occurrence quantities corresponding to the plurality of discrete values. Therefore, the calculation amount of the nodes in the Bayesian network when calculating the marginal probability is reduced, the calculation efficiency of the marginal probability is improved, and technical support is provided for meeting the marginal distribution calculation requirement in the massive data application scenario.

[0023] Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and / or can be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0024] The above and / or additional aspects and advantages of the application will become apparent and be more readily understood through consideration of the following description, taken in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A Bayesian network structure schematic diagram provided by an embodiment of the application;

[0026] Figure 2 A flowchart schematic diagram of an edge distribution acquisition method provided by an embodiment of the application;

[0027] Figure 3 An event occurrence quantity matrix schematic diagram provided by an embodiment of the application;

[0028] Figure 4 A Bayesian network training schematic diagram according to an embodiment of the application;

[0029] Figure 5 A flowchart schematic diagram of an edge distribution acquisition scenario provided by an embodiment of the application;

[0030] Figure 6 A structure schematic diagram of an edge distribution acquisition device provided by an embodiment of the application; and

[0031] Figure 7 A structure schematic diagram of another edge distribution acquisition device provided by an embodiment of the application. DETAILED DESCRIPTION

[0032] Embodiments of the application are described in detail below with reference to the attached drawings, which show by way of example, embodiments in which the same or similar elements are denoted by the same or similar reference numerals throughout the various figures. The embodiments described below are exemplary and are intended to explain the application, and should not be understood as limiting the application.

[0033] The edge distribution acquisition method, device, equipment and storage medium thereof of the embodiments of the present application are described below with reference to the accompanying drawings.

[0034] In the embodiments of the present application, the edge distribution acquisition refers to edge acquisition in a Bayesian network, which is a kind of probabilistic graphical model, a directed acyclic graph, determined by network structure and conditional probability distribution of each node, and defined by two elements of network structure and conditional probability distribution. In the process of implementing the embodiments of the present application, the inventors found that the edge distribution query based on a discrete Bayesian network can be performed using the method of joint probability distribution combined with the chain rule of Bayesian network, that is, in the Bayesian network, the edge distribution query based on a discrete Bayesian network is currently performed using the idea of joint probability distribution combined with the chain rule of Bayesian network. For the network structure in formula (1), the calculation method is as follows: P(A,B,C,D,E)=P(E|B)×P(B|C,A)×P(D|A)×P(C)×P(A). Figure 1

[0035] Alternatively, according to the satisfaction of the related requirements of the probability query by the exact inference algorithm, the common exact inference algorithms include variable elimination algorithm, clique tree algorithm, belief propagation algorithm, etc., wherein the variable elimination algorithm refers to the elimination of variables in the joint probability distribution by traversal summation when calculating the edge distribution, which uses the idea of dynamic programming.

[0036] The clique tree algorithm refers to using a clique tree to represent the joint probability distribution, and when using the clique tree algorithm to perform probability query on the Bayesian network, the Bayesian network needs to be converted into a clique tree structure (tree-shaped clustering graph) first and then the probability inference is performed. It is equivalent to the variable elimination algorithm in theoretical derivation, which converts the joint probability distribution into local factors, and then sums the product of the factors, while the clique tree algorithm converts the joint probability distribution into a clique tree using the variable elimination algorithm, and then performs information transmission between the clique trees.

[0037] The belief propagation algorithm refers to regarding the process of summing the probabilities of variables as information transmission, for each node, the probability distribution of the node is transmitted to the next node to change the probability distribution of the next node, and the edge distribution of each node is obtained by iterative convergence.

[0038] However, the inventors found that in the above edge distribution algorithms, after the structure learning and parameter learning of the discrete Bayesian network, only the conditional probability distribution of each node is retained as the parameter, and the original data is not retained, resulting in the need to traverse and search all nodes of the Bayesian network when querying the probability of a certain node, which is low in query efficiency, and when the network structure is large, it may cause low query efficiency or even unable to run.​

[0039] For example, when using variable elimination algorithms for inference queries in discrete Bayesian networks, finding the optimal elimination order is difficult. Variable elimination algorithms involve calculating a new vector while simultaneously eliminating variables. Assuming there are n variables, each with b different values, the algorithm's complexity is O(n²bk) when calculating a new vector with k variables. Furthermore, when multiple variable elimination operations are needed, intermediate results can often be reused, but the algorithm performs multiple calculations, resulting in significant redundant computation and low efficiency. In contrast, clustered tree algorithms only require information to be passed between leaf nodes and the root node, greatly reducing the computational scale, but they still cannot avoid traversing the entire graph to calculate the joint probability distribution. The belief propagation algorithm avoids the problem of redundant calculations caused by the variable elimination algorithm. The difference between it and the cluster tree algorithm is that the cluster tree algorithm first constructs the cluster tree and then propagates information on the cluster tree; while the belief propagation algorithm does not need to construct the cluster tree first and can perform calculations on the factor graph. It is more efficient than the cluster tree algorithm. However, the belief propagation algorithm is not applicable to arbitrary graphs. It only ensures convergence on the dendrogram, and the belief propagation algorithm still needs to traverse all nodes to complete the query of edge distribution.

[0040] To address the technical problems mentioned in the background section, this invention proposes an edge distribution acquisition method that focuses on the query efficiency of the edge distribution of a node in a Bayesian network. It skips the complex algorithms for joint probability distributions and the variable elimination algorithms with high computational complexity, using the number of event occurrences to query the edge distribution, thereby solving the problem of low timeliness in querying the marginal probability of Bayesian network nodes. The main idea is to eliminate the unnecessary calculations caused by the original technology's reliance on conditional distribution probabilities, and to add additional sample information to the probability information using the number of sample occurrences (support), simplifying the calculation and improving operational efficiency.

[0041] Figure 2 This is a flowchart illustrating a method for obtaining edge distributions according to an embodiment of the present invention. The edge distribution is a dimensionality reduction operation on the joint probability distribution, yielding a probability distribution with respect to only one variable. For example, for a correlation probability distribution P(x,y) with respect to two variables, the edge distribution with respect to node x is P(x) = ∑ y P(x,y)=∑ y P(x|y)P(y).

[0042] The conditional probability distribution is for two variables X and Y, the conditional probability distribution of variable Y under the condition {X=x} refers to the probability distribution of Y when variable X takes value x, i.e. P(Y|X=x) and the like. In addition, the Bayesian network mentioned in the embodiment refers to a probabilistic graphical model, which is a directed acyclic graph determined by the network structure and the conditional probability distribution of each node.

[0043] As shown in Figure 2 , the edge distribution acquisition method includes the following steps:

[0044] Step 101, acquiring an event occurrence number matrix corresponding to the conditional probability distribution of the target node of the Bayesian network.

[0045] Step 102, determining the event occurrence number corresponding to the plurality of discrete values of the target node and the total number of event occurrences of the target node according to the event occurrence number matrix corresponding to the conditional probability distribution.

[0046] Wherein, the target node can be understood as the node in the Bayesian network to be calculated for the edge distribution probability, wherein the plurality of discrete values of the target node can be understood as the probability value of the transmission occurrence probability between the other nodes associated with the target node and having a data transmission relationship, wherein the event corresponding to the plurality of discrete values can be understood as the transmission event between the other nodes associated with the target node and having a data transmission relationship, and the occurrence number of the event refers to the number of the event corresponding to the plurality of discrete values in the total number of events.

[0047] In the embodiment, the event occurrence number corresponding to the plurality of discrete values of the target node and the total number of event occurrences of the target node are acquired, and it is obvious that the probability of the event occurrence corresponding to each discrete value can be known based on the occurrence number and the total number of occurrences.

[0048] For example, referring to Figure 1 the Bayesian network structure diagram, when the target node is node D, if the conditional probability distribution P(D|A) is to be calculated, the discrete values acquired are various possibilities of P(D|A), such as P(D1|A1), P(D2|A2), and the like, the event occurrence number is the number of P(D1|A1) and the number of P(D2|A2), and the total number of event occurrences of the target node is the total number of all events occurring to all possible D and all possible A.

[0049] It should be noted that in different application scenarios, the way of acquiring the event occurrence number corresponding to the plurality of discrete values of the target node and the total number of event occurrences of the target node is different, and in the present example, the event occurrence number matrix corresponding to the conditional probability distribution of the target node is acquired.

[0050] In the embodiment, the event occurrence number matrix corresponding to the conditional probability distribution of the target node is stored in advance, the event occurrence number matrix includes the event occurrence number corresponding to each discrete value and the total event occurrence number of the target node.

[0051] In some possible embodiments, a large number of node identities and event occurrence matrices corresponding to each node identity can be stored in a preset application database, thereby, the application database corresponding to the target node can be acquired, the application database is queried according to the node identity of the target node, and the event occurrence number matrix corresponding to the conditional probability distribution of the target node is acquired.

[0052] Therefore, the event occurrence number corresponding to the target conditional probability distribution matching each discrete value is acquired from the event occurrence number matrix corresponding to the conditional probability distribution, and then the event occurrence number corresponding to the target conditional probability distribution is summed to acquire the event occurrence number corresponding to each discrete value.

[0053] Further, the event occurrence number corresponding to each discrete value of the target node and the total event occurrence number of the target node are determined according to the event occurrence number matrix corresponding to the conditional probability distribution.

[0054] In the embodiment, since the event occurrence number matrix includes the event occurrence number corresponding to each discrete value and the total event occurrence number of the target node, the event occurrence number corresponding to each discrete value of the target node and the total event occurrence number of the target node can be directly determined according to the event occurrence number matrix corresponding to the conditional probability distribution.

[0055] In the example, the event occurrence number matrix includes the occurrence number corresponding to each event, for example, as shown in Figure 3 The storage form of the conditional probability distribution P(D|A) when A and D take m different values is shown in the figure, and the event occurrence number matrix is an m*m matrix.

[0056] Of course, in some possible embodiments, the technology such as deep learning can also be used to obtain the sum of the event occurrence quantities corresponding to the plurality of discrete values to obtain the total event occurrence quantity corresponding to the plurality of discrete values. In this example, the deep model is trained in advance according to the deep learning technology based on the event occurrence probability of the sample node corresponding to each discrete value, so that the target node is input into the trained deep model to obtain the event occurrence probability of the target node corresponding to each discrete value. Based on the product of the preset event occurrence probability and the total quantity of all possible events in the corresponding scene, the event occurrence quantity corresponding to the plurality of discrete values of the target node is obtained, and then the sum of the event occurrence quantities corresponding to the plurality of discrete values is obtained to obtain the total event occurrence quantity corresponding to the plurality of discrete values. In step 103, the edge distribution of the target node is obtained based on the total event occurrence quantity and the event occurrence quantity corresponding to the plurality of discrete values by using the preset event quantity algorithm.

[0057] In this embodiment, the edge distribution of the target node is obtained based on the total event occurrence quantity and the event occurrence quantity corresponding to the plurality of discrete values.

[0058] In some possible examples, the ratio of the event occurrence quantity corresponding to each discrete value to the total event occurrence quantity is calculated, and the edge distribution of the target node is obtained based on the ratio corresponding to each discrete value.

[0059] For example, the edge distribution of the target node is calculated based on the conditional probability distribution of the target node and the quantity of the corresponding event occurrence. The edge distribution of the corresponding node is the quantity of the occurrence of each discrete value of the node divided by the total number of the event occurrence. For example, for the node D in the Bayesian network in Figure 1 , after P(D|A) and C(D,A) are obtained, wherein, C(D=m) is defined as the number of times of the event D=m. For a more complex conditional probability distribution, the same idea is used.

[0060] In some possible examples, the convolutional neural network is trained in advance, the total event occurrence quantity and the event occurrence quantity corresponding to the plurality of discrete values are input into the convolutional neural network, and the edge distribution of the target node is obtained. In some possible embodiments, the weight value corresponding to each discrete value is determined based on the experimental data, the product of the event occurrence quantity corresponding to the discrete value and the corresponding weight value is calculated, the ratio of the product value corresponding to the event occurrence quantity of the discrete value to the total event occurrence quantity is calculated, and the edge distribution of the target node is obtained.

[0061] Therefore, as shown in Figure 4 , in the edge distribution obtaining method of the embodiment of the present application, the Bayesian network is learned in advance, the quantity of each event occurrence is calculated and retained during the learning of the parameters of the Bayesian network, and the Bayesian network containing the quantity data is generated.

[0062] Thus, as Figure 5 shown, in calculating the edge probability of the Bayesian network, the number of occurrences of the time corresponding to each discrete value of the target node is obtained, and the total number of occurrences of the target node is obtained, and according to the total number of occurrences and the number of occurrences corresponding to the plurality of discrete values, the edge distribution of the target node is obtained.

[0063] Further, in an embodiment of the present application, the edge distribution of the target node can be obtained according to the corresponding query of the related event of the target node in the scene, such as querying the intelligence quotient of the target node, and the corresponding plurality of discrete values are the grades, education, etc. of the target node, and the obtained edge distribution can be used as the possible value of the intelligence quotient of the target node.

[0064] In summary, the edge distribution acquisition method of the embodiment of the present application improves the original parameter storage method of the Bayesian network in order to improve the query efficiency when querying the edge distribution of the discrete Bayesian network, and increases the number of occurrences of the event corresponding to the conditional probability distribution of each node. In addition, when querying the discrete Bayesian network, the original calculation method of the joint probability distribution combined with the variable elimination algorithm with higher complexity is abandoned, and the number of occurrences of the event is used for calculation, so that a more efficient node edge distribution query method is realized.

[0065] In order to realize the above-mentioned embodiment, the present application further provides an edge distribution acquisition device.

[0066] Figure 6 A structural schematic diagram of an edge distribution acquisition device provided by the embodiment of the present application.

[0067] As Figure 6 shown, the edge distribution acquisition device comprises a first acquisition module 710, a determination module 720 and a second acquisition module 730.

[0068] The first acquisition module 710 is configured to obtain an event occurrence number matrix corresponding to the conditional probability distribution of the target node of the Bayesian network.

[0069] The determination module 720 is configured to determine the number of occurrences of the event corresponding to the plurality of discrete values of the target node and the total number of occurrences of the target node according to the event occurrence number matrix corresponding to the conditional probability distribution.

[0070] The second acquisition module 730 is configured to obtain the edge distribution of the target node by using a preset event number algorithm according to the total number of occurrences and the number of occurrences corresponding to the plurality of discrete values. Further, in a possible implementation manner of the embodiment of the present application, as Figure 7 shown, the first acquisition module 710 comprises a first acquisition unit 711 and a determination unit 712.

[0071] The first obtaining unit 711 is configured to obtain an application database corresponding to the target node.

[0072] The determining unit 712 is configured to query the application database according to a node identifier of the target node, and obtain an event occurrence number matrix corresponding to a conditional probability distribution of the target node.

[0073] In a possible implementation of the embodiment of the application, the second obtaining module 730 includes a second obtaining unit 731 and a third obtaining unit 732, wherein,

[0074] The second obtaining unit 731 is configured to obtain, from the event occurrence number matrix corresponding to the conditional probability distribution, an event occurrence number corresponding to a target conditional probability distribution matched with each of the discrete values.

[0075] The third obtaining unit 732 is configured to sum the event occurrence numbers corresponding to the target conditional probability distribution, and obtain an event occurrence number corresponding to each of the discrete values.

[0076] In a possible implementation of the embodiment of the application, the second obtaining module 730 includes a calculating unit 733 and a fourth obtaining unit 734, wherein,

[0077] The calculating unit 733 is configured to calculate a ratio of the event occurrence number corresponding to each of the discrete values to the total event occurrence number.

[0078] The fourth obtaining unit 734 is configured to obtain an edge distribution of the target node according to the ratio corresponding to each of the discrete values.

[0079] It should be noted that the above explanation of the edge distribution obtaining method embodiment is also applicable to the edge distribution obtaining device of this embodiment, which will not be described here.

[0080] In order to implement the above-mentioned embodiments, the application further provides a computer device, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the edge distribution obtaining method described in the above-mentioned embodiments when executing the computer program.

[0081] In order to implement the above-mentioned embodiments, the application further provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the edge distribution obtaining method described in the above-mentioned embodiments.

[0082] In order to realize the above-mentioned embodiments, the application further provides a computer program product, which, when an instruction processor in the computer program product is executed, realizes the edge distribution acquisition method as described in the above-mentioned embodiments.

[0083] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples, without contradiction.

[0084] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the application, the meaning of "multiple" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0085] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing the specified logic functions or processes, and the preferred embodiments of the application also include additional implementation examples, in which the functions can be performed in different orders, in different ways, or in reverse, and the described embodiments should not be construed as limited to the described or discussed order or sequence of functions, unless otherwise specifically specified.

[0086] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of instructions to implement logic functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a machine-readable storage device (e.g., magnetic, optical or other) a machine-readable storage diskette (e.g., floppy, flexible or other), a machine-readable storage card (e.g., RAM, ROM or other), a machine-readable storage tape (e.g., magnetic, optical or other), a machine-readable storage medium (e.g., a portable memory chip), a machine-readable signal, a machine-readable propagated signal, a machine-readable compressed signal, and the like. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (e.g., a bus that has thin film resistors for

[0087] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the embodiments described above, various steps or methods can be implemented, in part, or in whole, in software / firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following techniques, or combinations thereof, can be used to implement the functions of the application: a discrete logic circuit(s) having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array(s) (PGA), a field programmable gate array (FPGA), and / or the like.

[0088] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0089] In addition, each function unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0090] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for edge distribution acquisition, said method being applied to mass image feature probability estimation, characterized in that, The method comprises the following steps: obtaining an event occurrence number matrix corresponding to a conditional probability distribution of a target node of a Bayesian network, the target node being an image whether meeting a first image feature; determining, according to the event occurrence number matrix corresponding to the conditional probability distribution, an event occurrence number corresponding to a plurality of discrete values of the target node and a total event occurrence number of the target node; obtaining, according to the total event occurrence number and the event occurrence numbers corresponding to the plurality of discrete values, an edge distribution of the target node by using a preset event number algorithm; wherein the step of obtaining the event occurrence number matrix corresponding to the conditional probability distribution of the target node of the Bayesian network comprises: obtaining an application database corresponding to the target node; querying the application database according to a node identifier of the target node to obtain the event occurrence number matrix corresponding to the conditional probability distribution of the target node.

2. The method of claim 1, wherein, The step of determining, according to the event occurrence number matrix corresponding to the conditional probability distribution, the event occurrence number corresponding to the plurality of discrete values of the target node comprises: obtaining, from the event occurrence number matrix corresponding to the conditional probability distribution, an event occurrence number corresponding to a target conditional probability distribution matched with each of the discrete values; summing the event occurrence numbers corresponding to the target conditional probability distributions to obtain an event occurrence number corresponding to each of the discrete values.

3. The method of claim 1, wherein, The step of obtaining, according to the total event occurrence number and the event occurrence numbers corresponding to the plurality of discrete values, the edge distribution of the target node by using the preset event number algorithm comprises: calculating a ratio of the event occurrence number corresponding to each of the discrete values to the total event occurrence number; obtaining the edge distribution of the target node according to the ratio corresponding to each of the discrete values.

4. An edge distribution acquisition apparatus, the apparatus being applied to a mass image feature probability estimation, characterized in that, The method comprises the following steps: a first obtaining module is configured to obtain an event occurrence number matrix corresponding to a conditional probability distribution of a target node of a Bayesian network, the target node being an image whether meeting a first image feature; a determining module is configured to determine, according to the event occurrence number matrix corresponding to the conditional probability distribution, an event occurrence number corresponding to a plurality of discrete values of the target node and a total event occurrence number of the target node; a second obtaining module is configured to obtain, according to the total event occurrence number and the event occurrence numbers corresponding to the plurality of discrete values, an edge distribution of the target node by using a preset event number algorithm. The first obtaining module comprises: a first obtaining unit is configured to obtain an application database corresponding to the target node, the application database storing a node identifier and an event occurrence matrix corresponding to each node identifier; a determining unit is configured to query the application database according to a node identifier of the target node to obtain the event occurrence number matrix corresponding to the conditional probability distribution of the target node.

5. The apparatus of claim 4, wherein, The second obtaining module comprises: a second obtaining unit is configured to obtain, from the event occurrence number matrix corresponding to the conditional probability distribution, an event occurrence number corresponding to a target conditional probability distribution matched with each of the discrete values; a third obtaining unit is configured to sum the event occurrence numbers corresponding to the target conditional probability distributions to obtain an event occurrence number corresponding to each of the discrete values.

6. The apparatus of claim 4, wherein, The second obtaining module comprises: a calculation unit configured to calculate a ratio of a number of events corresponding to each of the discrete values to a total number of events; a fourth acquisition unit configured to acquire the edge distribution of the target node according to the ratio corresponding to each of the discrete values.

7. A computer device, comprising: A computer program product, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the edge distribution acquisition method according to any one of claims 1-3 when executing the computer program.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program product is executed by the processor to implement the edge distribution acquisition method according to any one of claims 1-3.

Images