A data processing method, device, apparatus, and storage medium

Abstract

This application discloses a data processing method, apparatus, device, and storage medium, comprising: determining a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, wherein one scatter point of the first type represents an original labeled data; one scatter point of the second type represents unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents first data; based on the scatter plot, determining a neighborhood radius range of the first scatter point; within the neighborhood radius range, determining an extension region of the first scatter point; and configuring the labels of the second type scatter points within the extension region as the first labels of the first scatter point. The solution of this application consumes less time and storage space, and has high labeling accuracy.

Description

Technical Field

[0001] This application relates to the field of data processing technology, including but not limited to data processing methods, apparatus, devices, and storage media. Background Technology

[0002] With the continuous development of data processing technology, automated annotation technology has also been widely developed.

[0003] In related technologies, for image classification datasets with a small number of trustworthy labels, label propagation methods can be used to spread existing labels to other unlabeled data, thereby achieving automated labeling of the dataset. The label propagation algorithms in these technologies require constructing a relationship graph based on all data, calculating the probability transition matrix between each data point, and then labeling all datasets based on the probability transition matrix.

[0004] In related technologies, on the one hand, the computation of probability transition matrices consumes a lot of time and storage space, so it is time-consuming and requires a lot of storage space; on the other hand, the label propagation algorithm based on probability transition matrices will get a label for each data, so it cannot guarantee the accuracy of labeling.

[0005] Ensuring the accuracy of the dissemination scope and achieving efficient, simple, and easy-to-use label dissemination are problems that need to be solved. Summary of the Invention

[0006] This application provides a data processing method, apparatus, device, and storage medium.

[0007] The technical solution of this application is implemented as follows:

[0008] This application provides a data processing method, the method comprising: determining a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, wherein a scatter point of the first type represents an original labeled data; a scatter point of the second type represents unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents first data.

[0009] Based on the scatter plot, the neighborhood radius range of the first scatter point is determined; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate to.

[0010] Within the radius of the neighborhood, determine the extended region of the first scatter point;

[0011] The labels of the second type of scatter points within the extended region are configured as the first labels of the first scatter points; the first labels are used to characterize the data type to which the first data belongs.

[0012] This application provides a data processing apparatus, the apparatus comprising:

[0013] A first determining unit is configured to determine a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, one scatter point of the first type represents one original labeled data; one scatter point of the second type represents one unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents the first data.

[0014] The second determining unit is used to determine the neighborhood radius range of the first scatter point based on the scatter plot; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate.

[0015] The third determining unit is used to determine the extended region of the first scatter point within the neighborhood radius;

[0016] The processing unit is configured to assign the label of the second type of scatter points within the extended region to the first label of the first scatter point; the first label is used to characterize the data type to which the first data belongs.

[0017] This application also provides an electronic device, including: a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement the above-described data processing method.

[0018] This application also provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the above-described data processing method.

[0019] In this application's scheme, an extended region for the first scatter point is determined on the scatter plot, and the first label is propagated within this extended region. This approach offers several advantages: firstly, the entire process does not involve calculating the probability transition matrix, thus minimizing time and storage consumption; secondly, by propagating the label within the extended region, the method ensures label accuracy. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of an optional structure of the data processing system provided in an embodiment of this application;

[0021] Figure 2 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0022] Figure 3 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0023] Figure 4A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0024] Figure 5 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0025] Figure 6 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0026] Figure 7 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0027] Figure 8 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0028] Figure 9 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0029] Figure 10 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0030] Figure 11 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0031] Figure 12 This is a schematic diagram of an optional structure of the data processing apparatus provided in an embodiment of this application;

[0032] Figure 13 This is a schematic diagram of an optional structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of the application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0034] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0035] In the following description, the terms "first," "second," and "third" are used only to distinguish different objects and do not represent a specific order of objects, nor are they constituting a chronological order. It is understood that "first," "second," and "third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0037] This application provides data processing methods, apparatus, devices, and storage media. In practical applications, the data processing method can be implemented by a data processing apparatus, and the functional entities in the data processing apparatus can be collaboratively implemented by the hardware resources of electronic devices, such as computing resources like processors and communication resources (such as those used to support various communication methods like optical fiber and cellular networks).

[0038] The data processing method provided in this application is applied to a data processing system, which includes a data processing device and a database. The data processing device is used to annotate the data in the database.

[0039] Specifically, the data processing equipment is used to perform:

[0040] A first scatter point is determined on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, one scatter point of the first type represents an original labeled data; one scatter point of the second type represents unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents first data; based on the scatter plot, the neighborhood radius range of the first scatter point is determined; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate; within the neighborhood radius range, the extension region of the first scatter point is determined; the labels of the second type scatter points within the extension region are configured as the first labels of the first scatter point; the first label is used to characterize the data type to which the first data belongs.

[0041] It should be noted that the embodiments of this application do not limit the specific deployment method of the data processing equipment and database, and can be configured according to actual needs.

[0042] In one possible implementation, when the data processing device has sufficient storage space, the data processing device and the database can be integrated into the same electronic device.

[0043] In another possible implementation, when the storage space of the data processing device is small, the data processing device and the database can be deployed independently on different electronic devices.

[0044] As an example, data processing systems are applied to, for example... Figure 1 In the scenario shown, the data processing system 10 includes a data processing device 101 and a database 102. The data processing device 101 and the database 102 can communicate with each other.

[0045] Database 102 stores data to be labeled and labeled data.

[0046] Data processing device 101 is used to perform the following operations on data in database 102: Determine a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, one scatter point of the first type representing an original labeled data point; one scatter point of the second type representing an unlabeled data point; the first scatter point is any scatter point of the first type; the first scatter point represents first data; based on the scatter plot, determine the neighborhood radius range of the first scatter point; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius to which the first scatter point can propagate; within the neighborhood radius range, determine the extended region of the first scatter point; configure the labels of the second type scatter points within the extended region as the first labels of the first scatter point; the first label is used to characterize the data type to which the first data belongs.

[0047] The labeled data will then be sent to database 102.

[0048] The data processing device 101 can be an electronic device with relevant data processing capabilities. For example, the data processing device 101 can be a server.

[0049] Database 102 can be an electronic device with relevant data storage capabilities. For example, database 102 can be a storage device.

[0050] Below, in conjunction with Figure 1 The schematic diagram of the data processing system shown illustrates various embodiments of the data processing method, apparatus, device, and storage medium provided in this application.

[0051] In a first aspect, embodiments of this application provide a data processing method, which is applied to a data processing apparatus; wherein the data processing apparatus can be deployed in... Figure 1 The data processing device 101 is used in the process described below. The data processing procedure provided in this embodiment will be explained using the data processing device 101 as the executing entity.

[0052] Figure 2The diagram illustrates an optional data processing method for labeling data using a label propagation approach.

[0053] Below, we will use the first scatter point as an example to explain the label propagation process. The label propagation process for other scatter points is similar to that for the first scatter point. For the specific implementation process, please refer to the detailed description for the first scatter point, which will not be repeated here.

[0054] refer to Figure 2 The label propagation process of the first scatter point, as shown in the diagram, may include, but is not limited to, the following: Figure 2 S201 to S204 are shown.

[0055] S201. The data processing device determines the first scatter point on the scatter plot.

[0056] Scatter plots are used to display data in a visual way.

[0057] In a scatter plot, each point represents a data point; the distance between points represents the similarity between the data represented by the scatter points, or the feature similarity between the data represented by the scatter points. Simply put, the higher the similarity between two data points, the smaller the distance between their corresponding points in the scatter plot.

[0058] The data here can be image data, text data, audio data, etc. This application does not limit the specific type of data; it can be determined according to the actual situation.

[0059] Specifically, the scatter plot in this embodiment includes: a first type of scatter plot and a second type of scatter plot. A first type of scatter plot represents a piece of original labeled data, also known as trusted labeled data; a second type of scatter plot represents unlabeled data.

[0060] Here, the original labeled data for the first type of scatter plot representation can be manually labeled, trustworthy labeled data. The unlabeled data for the second type of scatter plot representation is the data to be labeled.

[0061] The first scatter point is any scatter point of the first type. Here, the first scatter point represents the first data point.

[0062] S201 can be implemented as follows: The data processing device searches for a first type of scatter point on the scatter plot, and arbitrarily determines one of the three points of the first type as the first scatter point to be processed.

[0063] S202. The data processing device determines the neighborhood radius range of the first scatter point based on the scatter plot.

[0064] The neighborhood radius range is used to characterize the range of values ​​for the neighborhood radius that the first scatter point can propagate to.

[0065] The embodiments of this application do not limit the specific method for determining the neighborhood radius range of the first scatter point, and can be configured according to actual needs.

[0066] In one possible implementation, the data processing device can determine the range of the neighborhood radius of the first scatter point based on the maximum value of the neighborhood radius of the first scatter point and the minimum value of the neighborhood radius of the first scatter point.

[0067] In another possible implementation, the data processing device may determine the range of the neighborhood radius of the first scatter point based on the maximum value of the neighborhood radius of the plurality of second scatter points and the minimum value of the neighborhood radius of the plurality of second scatter points.

[0068] In this possible implementation, in the scatter plot, the label of the first scatter point is a first label, and there is more than one scatter point corresponding to the first label; that is, there are multiple second scatter points corresponding to the first label. Therefore, it can be understood that these multiple second scatter points should have the same neighborhood radius range. That is, the neighborhood radius range of the first scatter point can be determined based on the maximum value of the neighborhood radius of the multiple second scatter points and the minimum value of the neighborhood radius of the multiple second scatter points.

[0069] S203. The data processing device determines the extended area of ​​the first scatter point within the radius of the neighborhood.

[0070] For example, when the extended region of the first scatter point is circular, the extended radius corresponding to the extended region of the first scatter point is within the radius of the neighborhood.

[0071] S203 can be implemented as follows: within the neighborhood radius, the data processing device starts from the minimum value of the neighborhood radius and gradually expands outwards, obtains the expansion radius of the first scatter point according to the expansion rule, and determines the expansion area of ​​the first scatter point based on the first scatter point and the expansion radius of the first scatter point.

[0072] This application does not impose specific limitations on the specific extension rules, and can be configured according to actual needs.

[0073] For example, the extended region of the first scatter point cannot intersect with the extended regions of other first-type scatter points.

[0074] For example, the extended region of the first scatter point does not include other first-type scatter points.

[0075] Determining the extended region of the first scatter point based on the first scatter point and its extended radius can be implemented as follows: the extended region of the first scatter point is determined to be a circular region with the first scatter point as the center and the extended radius as the radius.

[0076] It is understandable that the extended region of the first scatter point can also be a square, which will not be elaborated on here.

[0077] S204. The data processing device configures the labels of the second type of scatter points in the extended area as the first labels of the first scatter points.

[0078] The first label is used to identify the data type to which the first data belongs.

[0079] It is understandable that the second type of scatter points within the extended area of ​​the first scatter point can be considered to have the same label as the first scatter point. Therefore, S204 can be implemented as follows: the data processing device configures the label of the second type of scatter points within the extended area as the first label of the first scatter point.

[0080] This application does not limit the specific method of configuring the labels of scattered points; it can be determined according to the actual situation.

[0081] In one possible implementation, the labels for the scatter points can be configured based on a table.

[0082] For example, the labels for scatter points can be as shown in Table 1 below.

[0083] Table 1 Examples of scatter plot labels

[0084] Scatter Scatter 1 Scatter 2 Scatter 3 …… Label Tag 1 Tag 2 Tag 1 ……

[0085] In another possible implementation, labels can be directly placed on the scatter plot. In this example, the labels can be in the form of text; or, different identifiers or colors can be configured for different labels, and the corresponding identifiers or colors can be used to mark the relevant locations on the scatter plot.

[0086] This application provides a data processing method, comprising: determining a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, wherein a scatter point of the first type represents an original labeled data; a scatter point of the second type represents unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents first data; based on the scatter plot, determining the neighborhood radius range of the first scatter point; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius to which the first scatter point can propagate; within the neighborhood radius range, determining an extension region of the first scatter point; configuring the labels of the second type scatter points within the extension region as the first label of the first scatter point; the first label is used to characterize the data type to which the first data belongs. In this application, the extension region of the first scatter point is determined on the scatter plot, and the first label is propagated within the extension region. Thus, on the one hand, the entire processing does not involve the calculation of a probability transition matrix, so the time and storage space consumed are small; on the other hand, the method of this application only propagates labels within the extension region, although some data may not be labeled, the accuracy of the labeling can be guaranteed.

[0087] The process by which the S202 data processing device determines the neighborhood radius of the first scatter point based on the scatter plot will be described below. This process may include, but is not limited to, the following two methods:

[0088] Method 1: For a scatter plot that includes a second scatter point, determine the neighborhood radius of the first scatter point;

[0089] Method 2: For a scatter plot that includes N second scatter points, determine the neighborhood radius range of the first scatter point.

[0090] The label of the second scatter point is the first label.

[0091] The following describes the process of determining the neighborhood radius of the first scatter point in Method 1, where the scatter plot includes a second scatter point, i.e., the second scatter point refers to the aforementioned first scatter point.

[0092] like Figure 3 As shown, the process may include, but is not limited to, S301 to S305 below.

[0093] S301, The data processing device determines a third scatter point on the scatter plot.

[0094] The third scatter point is a second-type scatter point, and the first distance between the third scatter point and the second scatter point is less than the distance between the second scatter point and all other scatter points on the scatter plot except the third scatter point. Simply put, the third scatter point is the scatter point on the scatter plot that is closest to the second scatter point.

[0095] S301 can be implemented as follows: the data processing device calculates the distance between each nearby scatter point and the second scatter point on the scatter plot, and determines the scatter point with the smallest distance from the second scatter point as the third scatter point; the distance between the third scatter point and the second scatter point is taken as the first distance.

[0096] S302, The data processing device determines the first distance as the starting value of the neighborhood radius range of the second scatter point.

[0097] The data processing device determines the initial value of the neighborhood radius range of the second scatter point as the first distance obtained in S301.

[0098] S303, The data processing device determines L fourth scatter points on the scatter plot.

[0099] The fourth scatter point is located inside the circle obtained with the second scatter point as the center and a preset value as the radius.

[0100] L is an integer greater than 1.

[0101] Simply put, the fourth scatter point is the scatter point near the second scatter point on the scatter plot.

[0102] This application embodiment does not limit the number of L or the size of the preset value, and can be configured according to actual needs. Specifically, on the scatter plot, the smaller the preset value, the smaller the number of L.

[0103] S304. The data processing device determines a second distance between every two of the L fourth scattered points.

[0104] The data processing device calculates the second distance between every two fourth scattered points in the L fourth scattered points, and obtains multiple second distances.

[0105] S305. The data processing device determines the maximum second distance value among the second distances between every two fourth scattered points as the endpoint value of the neighborhood radius range of the second scattered point.

[0106] The data processing device determines the maximum second distance value among the multiple second distances obtained in S304, and sets the maximum second distance value as the endpoint value of the neighborhood radius range of the second scatter point.

[0107] The following section explains the process of determining the neighborhood radius of the first scatter point in Method 2, where the scatter plot includes N second scatter points (i.e., the second scatter points also include other scatter points besides the first scatter point). N is an integer greater than 1.

[0108] like Figure 4 As shown, the process may include, but is not limited to, S401 to S404 below.

[0109] S401. For each of the N second scattered points, the data processing device determines the initial value of the neighborhood radius of each second scattered point, thus obtaining N initial values.

[0110] The process of determining the initial value of the neighborhood radius of each second scatter point can be found in the detailed descriptions of S301 and S302, which will not be repeated here.

[0111] By referring to the process of determining the initial value of each neighborhood radius, N initial values ​​are obtained.

[0112] S402. The data processing device determines the smallest starting value among the N starting values ​​as the starting value of the neighborhood radius range of the first scatter point.

[0113] The data processing device finds the smallest starting value among N starting values ​​and determines the starting value of the radius range of the first scatter point as that smallest starting value.

[0114] S403. For each of the N second scattered points, the data processing device determines the endpoint value of the neighborhood radius of each second scattered point, and obtains N endpoint values.

[0115] The data processing device refers to the process from S303 to S305 to obtain the endpoint value of the neighborhood radius of each second scatter point, thereby obtaining N endpoint values.

[0116] S404. The data processing device determines the minimum endpoint value among the N endpoint values ​​as the endpoint value of the neighborhood radius range of the first scatter point.

[0117] The data processing device determines the smallest endpoint value among N endpoint values, and determines the endpoint value of the neighborhood radius range of the first scatter point as the smallest endpoint value.

[0118] The process by which the S203 data processing device determines the extended region of the first scatter point within the neighborhood radius will be described below. This process may include, but is not limited to, the following two methods:

[0119] Method A: Determine the expansion region of the first scatter point based on the expansion rules;

[0120] Method B: Determine the expansion region of the first scatter point based on the expansion rules, and adjust the expansion region of the first scatter point.

[0121] The following describes the process of determining the extended region of the first scatter point based on the extended rules in method A.

[0122] like Figure 5 As shown, the process may include, but is not limited to, S501 to S504 below.

[0123] S501, The data processing device determines the neighborhood radius of the i-th round.

[0124] The neighborhood radius of the i-th round is within the range of the neighborhood radius; i is an integer greater than or equal to 1.

[0125] In the first round, the value of i is 1. As the round expands, i increases by 1 with each expansion.

[0126] When i equals 1, the neighborhood radius of the first round is the minimum value of the neighborhood radius range of the first scatter point. As the expansion continues and the number of rounds increases, the neighborhood radius of each round will also change. However, the neighborhood radius of each round is always within the range of the neighborhood radius.

[0127] S502. The data processing device takes the first scattered point as the center and the neighborhood radius of the i-th round as the expansion radius, and expands outward based on the first step length to obtain the expansion result of the i-th round.

[0128] The expansion result of round i is used to indicate that the expansion radius of the first scatter point in round i is the neighborhood radius of round i.

[0129] In one possible implementation, the data processing device expands outward from the first scatter point as the center, with the neighborhood radius of the i-th round as the expansion radius and the first step length as a fixed length, to obtain the expansion result of the i-th round.

[0130] In another possible implementation, the data processing device expands outward with the first scatter point as the center, the neighborhood radius of the i-th round as the expansion radius, and the first step length as the initial step length, by gradually increasing or decreasing the step length, to obtain the expansion result of the i-th round.

[0131] In this embodiment, the specific length of the first step is not limited and can be configured according to actual needs.

[0132] For example, the first step length can be the minimum number of neighborhood points (MinPts).

[0133] S503. The data processing device determines the expansion radius of the first scatter point based on the expansion result of the i-th round.

[0134] The expansion radius of the first scatter point is used to characterize the maximum expansion radius of the first scatter point in multiple rounds of expansion when expansion is successful.

[0135] The process and rules for determining the expansion radius of the first scatter point based on the expansion results of the i-th round are not uniquely limited in this application embodiment, and can be configured according to actual needs.

[0136] S504. The data processing device determines that the extended region of the first scatter point is a circular region with the first scatter point as the center and the extended radius as the radius.

[0137] The data processing device determines the extended region of the first scatter point as a circular region with the first scatter point as the center and the extended radius as the radius.

[0138] The process by which the S503 data processing device determines the expansion radius of the first scatter point based on the expansion results of the i-th round will be described below. This process may include, but is not limited to, Case 1 or Case 2 below.

[0139] For case 1, such as Figure 6 As shown, the process of determining the expansion radius of the first scatter point may include, but is not limited to, S601 and S602 below.

[0140] S601. If the first expansion region of the i-th round includes the fifth scatter point, the data processing device cancels the expansion result of the i-th round, reduces the neighborhood radius of the (i+1)-th round, and determines the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)-th round.

[0141] The label of the fifth scatter point is the second label, and the fifth scatter point is a scatter point of the first type.

[0142] The first expansion region in the i-th round is used to characterize the newly added region in the i-th round of expansion during the first scatter expansion.

[0143] S601 can be implemented as follows: if the first expansion region in the i-th round includes the fifth scatter point, then the data processing device cancels the expansion result of the i-th round, that is, cancels the expansion region newly added to the first scatter point in the i-th round. Then, the neighborhood radius of the (i+1)-th round is reduced, and the expansion radius of the first scatter point is determined based on the reduced neighborhood radius of the (i+1)-th round.

[0144] The number of fifth scattered points included in the first expansion region of the i-th round in this application embodiment is not limited; it can be one or more.

[0145] In the process of reducing the neighborhood radius of the (i+1)th round and determining the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)th round, the step size and method of reduction are not limited and can be configured according to actual needs.

[0146] S602. If the first expansion region does not include the fifth scatter point, the data processing device increases the neighborhood radius of the (i+1)th round, takes the (i+1)th round as the new i-th round, and continues to expand until the first expansion region of the new i-th round includes the fifth scatter point.

[0147] S602 can be implemented as follows: if the first expansion region does not include the fifth scatter point, the data processing device increases the neighborhood radius of the (i+1)th round, takes the (i+1)th round as the new i-th round, and continues to expand until the first expansion region of the new i-th round includes the fifth scatter point. Then, the expansion result of the new i-th round is canceled, and the neighborhood radius of the new (i+1)th round is reduced. The expansion radius of the first scatter point is determined based on the reduced neighborhood radius of the new (i+1)th round.

[0148] In this application, the specific implementation of increasing the neighborhood radius in the (i+1)th round is not limited, and can be configured according to actual needs. For example, it can be increased by a fixed value or gradually.

[0149] For case 2, such as Figure 7 As shown, the process of determining the expansion radius of the first scatter point may include, but is not limited to, S701 and S702 below.

[0150] S701. If the first expansion region corresponding to the second data occupied by the i-th round intersects with the second expansion region corresponding to the second data occupied by the sixth scatter point, the data processing device determines the expansion radius of the first scatter point based on the size relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data.

[0151] Specifically, the newly added expansion area corresponding to the second data occupied by the first scatter point in the i-th round is taken as the first expansion area, and the newly added area corresponding to the second data occupied by the sixth scatter point is taken as the second expansion area. If the first expansion area and the second expansion area have an intersection, the data processing device determines the expansion radius of the first scatter point based on the size relationship between the neighborhood radius in the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data.

[0152] The sixth scatter point is a first-type scatter point, and its label is different from that of the first scatter point.

[0153] The embodiments of this application do not limit the process by which the data processing device determines the expansion radius of the first scatter point based on the relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data. The configuration can be made according to actual needs.

[0154] S702. If the first expansion region and the second expansion region do not intersect, the data processing device increases the neighborhood radius of the (i+1)th round, takes the (i+1)th round as the new i-th round, and continues to expand until the first expansion region corresponding to the second data in the new i-th round intersects with the second expansion region when the sixth scatter point occupies the second data.

[0155] This application does not limit the method of increasing the neighborhood radius in the (i+1)th round; it can be configured according to actual needs. For example, it can be increased by a fixed value or by a gradual increase.

[0156] If the first expansion region and the second expansion region do not intersect, the data processing device increases the neighborhood radius of the (i+1)th round, and takes the (i+1)th round as the new i-th round, and continues to expand until the first expansion region corresponding to the second data in the new i-th round intersects with the second expansion region when the sixth scatter point occupies the second data. Based on the size relationship between the neighborhood radius of the new i-th round and the second neighborhood radius when the sixth scatter point occupies the second data, the expansion radius of the first scatter point is determined.

[0157] The following describes the process of determining the expansion radius of the first scatter point based on the relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data.

[0158] If the neighborhood radius of the i-th round is greater than or equal to the second neighborhood radius, the data processing device cancels the expansion result of the i-th round and reduces the neighborhood radius of the (i+1)-th round, and determines the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)-th round.

[0159] The data processing device cancels the expansion result of the i-th round, which means canceling the newly added expansion area of ​​the i-th round and canceling the expansion radius of the i-th round.

[0160] The embodiments of this application do not limit the method of reducing the neighborhood radius of the (i+1)th round or the process of determining the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)th round, and can be configured according to actual needs.

[0161] For example, the decrease can be made by a fixed value or by a gradual decrease.

[0162] If the neighborhood radius of the i-th round is smaller than the second neighborhood radius, the data processing device increases the neighborhood radius of the (i+1)-th round, and uses the (i+1)-th round as the new i-th round, and continues to expand.

[0163] The embodiments of this application do not limit the specific implementation of increasing the neighborhood radius in the (i+1)th round, and can be configured according to actual needs. For example, it can be increased by a fixed value, or it can be increased gradually.

[0164] The process of determining the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)th round is explained below.

[0165] If the neighborhood radius of the (i+1)th round is less than or equal to the neighborhood radius of the previous round, then the neighborhood radius of the previous round is determined as the expansion radius of the first scatter point; if the neighborhood radius of the (i+1)th round is greater than the neighborhood radius of the previous round, then the neighborhood radius of the (i+2)th round is increased, and the (i+2)th round is taken as the new (i+1)th round, and expansion continues until the neighborhood radius of the new (i+1)th round is less than or equal to the neighborhood radius of the previous round, and the neighborhood radius of the previous round of the new (i+1)th round is determined as the expansion radius of the first scatter point.

[0166] Method B: Determine the expansion region of the first scatter point based on the expansion rules, and adjust the expansion region of the first scatter point.

[0167] The following describes the process of determining the expansion region of the first scatter point based on the expansion rule in Method B, and adjusting the expansion region of the first scatter point.

[0168] In Method B, the process of determining the extended region of the first scatter point based on the extended rules is the same as that of Method A. For details, please refer to the detailed description of Method A. It will not be repeated here.

[0169] The following explains the adjustment process for the expanded region of the first scatter point. For example... Figure 8 As shown, the process may include, but is not limited to, S801 and S802.

[0170] S801. The data processing device calculates the third distance between the first scattered point and the M seventh scattered points to obtain the M third distances.

[0171] Where M is an integer greater than 0.

[0172] S802. If H of the M third distances and the neighborhood radius of the extended region satisfy a non-specified condition, the data processing device receives a first operation and adjusts the extended region of the first scatter point based on the first operation.

[0173] H is less than or equal to M.

[0174] Among them, the unspecified conditions correspond to the specified relationships.

[0175] Example 1: The specified relationship indicates that the seventh point should not originally belong to the actual extended region of the first point. The unspecified condition indicates that the third distance is less than the neighborhood radius, that is, the extended region of the first point includes the seventh point.

[0176] S802 can be implemented as follows: the data processing device receives a first operation and adjusts the expansion area of ​​the first scatter point based on the first operation so that the seventh scatter point corresponding to the H third distances is located outside the expansion area of ​​the first scatter point.

[0177] The first operation can be a click or touch operation on the scatter plot. The data processing device reduces the expansion radius of the first scatter plot based on the location of the click or touch operation, and determines the expansion area of ​​the first scatter plot based on the reduced expansion radius.

[0178] Example 2: The specified relationship indicates that the seventh scatter point should have belonged to the actual extended region of the first scatter point. The unspecified condition indicates that the third distance is greater than the neighborhood radius, that is, the extended region of the first scatter point does not include the seventh scatter point.

[0179] S802 can be implemented as follows: the data processing device receives the first operation and increases the expansion area of ​​the first scatter point based on the first operation, so that the seventh scatter point corresponding to the H third distances is located within the expansion area of ​​the first scatter point.

[0180] The first operation can be a click or touch operation on the scatter plot. The data processing device increases the expansion radius of the first scatter plot based on the location of the click or touch operation, and determines the expansion area of ​​the first scatter plot based on the increased expansion radius.

[0181] Furthermore, the data processing method provided in this application embodiment can also display the extended region of the first scatter point. Specifically, the data processing device displays the extended region of the first scatter point at the first position corresponding to the first scatter point shown on the scatter plot.

[0182] In this way, the expanded area of ​​each scatter point can be visually observed on the scatter plot.

[0183] Furthermore, the data processing method provided in this application embodiment can also visualize the scatter points within the extended area of ​​the first scatter point. Specifically, the data processing device displays the first scatter point and the second type of scatter points in the same form within the extended area corresponding to the first scatter point on the scatter plot. This same form can be the same scatter point color, scatter point shape, or the same labeling method, etc.

[0184] The following example, using image data, illustrates the data processing method provided in this application.

[0185] In related technologies, for image classification datasets with a small number of trustworthy labels, label propagation methods can be used to spread existing labels to other unlabeled data, thereby achieving automated labeling of the dataset. However, label propagation algorithms in these technologies require constructing a relationship graph based on all data and calculating the probability transition matrix between each data point. The entire process is time-consuming and storage-intensive, and the resulting labeling results are unstable.

[0186] Ensuring the accuracy of the dissemination scope and achieving efficient, simple, and easy-to-use label dissemination are problems that need to be solved.

[0187] To address the aforementioned problems, the data processing method provided in this embodiment of the application can adaptively adjust the propagation range based on feature distance and achieve visualized label propagation. Specifically:

[0188] 1. Extract, standardize, and select data features.

[0189] 2. Reduce the dimensionality of the dataset, visualize the dataset, and highlight the data with trustworthy labels.

[0190] 3. Using each trusted label data as the core object, and an adaptively adjustable neighborhood semidiameter (Eps) as the neighborhood radius, with a fixed value of 1 as the minimum number of neighborhood points (MinPts), a density-based clustering algorithm (Density-Based Spatial Clustering of Applications with Noise, DBSCAN) is used to expand the density-reachable region of each trusted label data round by round. Ultimately, all points connected to the density of each trusted label data constitute the expanded range of each trusted label data. Specifically, this may include, but is not limited to, steps 1) and 2) below.

[0191] Step 1) Based on the distribution of data near each trusted label data after dimensionality reduction, calculate the range of values ​​for the neighborhood radius (Eps) of each trusted label data [Eps-min, Eps-max].

[0192] Step 2) For each trusted label data, multiple rounds of expansion can be carried out simultaneously. During the expansion process, the neighborhood radius of each trusted label data can be adaptively adjusted, and the expansion results will be iterated in each round. If they meet during the expansion process, the data will be assigned according to the rules, which ensure the stability and accuracy of the expansion results.

[0193] Specific extension rules may include, but are not limited to, the following rules A through E.

[0194] Rule A: During the expansion process, if trusted label data with different labels is encountered, the expansion is stopped and the results of this round of expansion are canceled. The neighborhood radius is reduced to a new neighborhood radius (Eps-new) based on the neighborhood radius determined in this round. If Eps-new is less than Eps in the previous round, the expansion is terminated; otherwise, a new round of expansion begins with Eps-new.

[0195] Rule B: During the expansion process of the trustworthy labeled data A, if the data p to be occupied has already been occupied by the trustworthy labeled data B, then compare the neighborhood radii of A and B. If the neighborhood radius of A (Eps-A) is greater than the neighborhood radius of p when it was occupied by B (Eps-p), that is, Eps-A > Eps-p, then A stops expanding, cancels the expansion result of this round, and performs the same operation as Rule A. If Eps-A < Eps-p, then the data p belongs to the expansion range of A, and A continues to expand; for B, B cancels all expansion results after Eps-p and performs the same operation as Rule A. If Eps-A = Eps-p, then A performs the same operation as Rule A, and B cancels all expansion results after Eps-p.

[0196] Rule C: If after a round of expansion, the Eps value of the trustworthy labeled data is not reduced, then increase the Eps value of this trustworthy labeled data to Eps-new according to the rule. If Eps-new is greater than Eps-max, then the expansion terminates; otherwise, use Eps-new to calculate the new expansion range for a new round.

[0197] Rule D: For trustworthy labeled data with the same label, their Eps value ranges are the minimum of their respective Eps value ranges. They expand separately centered on themselves, and finally, the expansion ranges are merged as the expansion range of this label.

[0198] Rule E: During the expansion process, if data occupied by the same label is encountered, it is ignored.

[0199] 4. Visually view the expansion results and adjust the neighborhood radii of each trustworthy labeled data. Among them, this process can include but is not limited to the following steps 01) to step 04).

[0200] Step 01): During the expansion process, the expansion results can be directly viewed visually. Among them, the expanded data has the same color and scatter point shape as its corresponding trustworthy labeled data. When the mouse hovers over the expanded data, the corresponding label and the neighborhood radius when the data was expanded will be displayed. The scatter points of the data expanded in different rounds have different sizes, and the scatter points in earlier rounds are larger.

[0201] Step 02): After the expansion stops, display the data pictures of the expansion edges of each trustworthy labeled data.

[0202] Step 03): Measure the characteristic distances between data points using a visual distance measurement tool.

[0203] Step 04): Adjust the neighborhood radii of the trustworthy labeled data as needed.

[0204] 5. Assign the labels of each trustworthy labeled data to all data within their respective expansion ranges.

[0205] This embodiment of the application has the following effects: by adaptively adjusting the neighborhood radius of each trusted label data, accurate, stable, and efficient calculation of the expansion range of each trusted label data is achieved; by visualizing the expansion results of each trusted label data and a visualized feature distance measurement tool, subjective initiative is leveraged to further ensure the accuracy and reliability of the expansion range. Ultimately, this enables each trusted label data to be accurately, efficiently, stably, and conveniently disseminated to other unlabeled data.

[0206] refer to Figure 9 The main workflow shown may include, but is not limited to, S901 to S909 below.

[0207] S901. Acquire and process features. Specifically, such as... Figure 9 As shown, acquiring and processing features may include: acquiring features and performing processes such as numericalization, standardization, normalization, and feature selection.

[0208] S902. The features are reduced in dimensionality using the Uniform Manifold Approximation and Projection (UMAP) algorithm.

[0209] S903. Draw a scatter plot to visualize the dataset.

[0210] S904. Calculate the neighborhood radius range of each trusted label data.

[0211] Specific rules and procedures may include, but are not limited to: Figure 10 S1001 to S1014 are shown.

[0212] S1001. Obtain the trusted tag data set C.

[0213] S1002. Determine if C is empty.

[0214] If C is empty, the process ends; if C is not empty, then execute S1003 below.

[0215] S1003. Extract the set of trusted label data D corresponding to a label L.

[0216] S1004. Determine whether D contains only one data item.

[0217] If D contains only one data A, then execute S1005 below; if D contains more than one data A, then execute S1011 below.

[0218] S1005, Find the data M that is closest to the trusted tag data A.

[0219] S1006. Determine whether M is trustworthy tag data.

[0220] If M is trusted label data, then execute S1010; if M is not trusted label data, then execute S1007 as follows.

[0221] S1007. The feature distance between A and M is taken as the minimum neighborhood radius of A, Eps-min.

[0222] S1008. Take the maximum value of the pairwise nearest feature distances between N points near A as the maximum neighborhood radius Eps-max of A.

[0223] S1009. Place A (including its neighborhood radius range information) into the calculation set W to be expanded.

[0224] The set of computations to be expanded, W, can also be called the set of label data to be expanded, W.

[0225] S1010. Set the radius range of A, Eps-min and Eps-max, to 0; mark A as completed in extended computation and add A to the set F of completed extended computation.

[0226] S1011. Extract all data A1, A1...An from D.

[0227] S1012. Find the data Mn that is closest to each element in D. Mn does not belong to D. Calculate the Eps-min and Eps-max of each element in D in a similar way to A.

[0228] For specific implementation details, please refer to S1007 and S1008.

[0229] S1013. Take the minimum Eps-min and Eps-max of each element in D as the neighborhood radius range of all elements in D, that is, the label L.

[0230] S1014. Place each element in D (including its neighborhood radius range information) into the set W to be expanded for computation.

[0231] After executing S1009, S1010, and S1014, S1002 is executed again to retrieve another label and perform a similar operation.

[0232] S905, Highlight trusted label data.

[0233] Assign different scatter plot shapes and colors to trusted tag data with different labels, and highlight them. When the mouse hovers over them, the label, data image, and neighborhood radius range are displayed.

[0234] S906. Taking each trusted tag data as the center, calculate the expansion range of each trusted tag data.

[0235] refer to Figure 11 The content shown may include, but is not limited to, S1101 to S1114 below, the specific extended process.

[0236] S1101. Obtain the data set W of the tags to be expanded.

[0237] S1102. Determine if W is empty.

[0238] If W is empty, then the process ends; if W is not empty, then execute S1103 below.

[0239] S1103. Extract a trustworthy label data A. Domain radius Eps = Eps-min, Eps-ok = Eps-min.

[0240] Here, Eps represents the current neighborhood radius, and Eps-OK represents the neighborhood radius after a successful expansion.

[0241] S1104. Determine whether Eps is greater than Eps-max.

[0242] If Eps is greater than Eps-max, then execute S1105 below; if Eps is not greater than Eps-max, then execute S1106 below.

[0243] S1105. The extended range calculation of A is completed. Place A into the set F where the extended range calculation is completed.

[0244] After S1105 is executed, S1102 is executed to retrieve another trusted tag data and perform the same processing.

[0245] S1106. Using Eps as the neighborhood radius and 1 as minPts, expand to the density-accessible region.

[0246] S1107. Determine whether other trustworthy tag data with different tags are encountered during expansion.

[0247] If yes, then execute S1108 below; if no, then execute S1110 below.

[0248] S1108, 1) Cancel the results of this round of expansion; 2) Reduce the value of A to Eps-new by a fixed step size, and set Eps-max to Eps-new.

[0249] S1109. Determine if Eps-new is greater than Eps-OK.

[0250] If yes, then execute S1106 again; otherwise, execute S1105.

[0251] S1110. Determine whether data P occupied by other different labels (B) is encountered during expansion.

[0252] If yes, execute S1111; otherwise, execute S1114.

[0253] S1111 Determine whether Eps is greater than Eps-p.

[0254] If yes, then execute S1108; otherwise, execute S1112 as described below.

[0255] S1112, 1) Cancel the labeling results of B in all rounds where the neighborhood radius is greater than or equal to Eps-p; 2) Delete Eps-p and all subsequent values ​​in B's Eps-ok list.

[0256] S1113. Determine whether Eps is equal to Eps-p.

[0257] If yes, then execute S1108; otherwise, execute S1114.

[0258] S1114, 1) is the data record label and Eps value for this round of expansion; 2) according to a fixed step size, increase A's Eps to Eps-new, set Eps-ok = Eps, Eps = Eps-new; 3) add Eps-ok to A's Eps-ok list.

[0259] After executing S1114, execute S1104 again to proceed with the next round of expansion.

[0260] S907. Identify the extended range of each trusted label's data and display a data image of the extended edge.

[0261] Hovering the mouse over any data within the extended range will display the extended label and the radius of the neighborhood extended to that data.

[0262] Furthermore, the maximum value of the neighborhood radius of trusted label data can be visually adjusted as needed, and the expanded range can be recalculated.

[0263] S908. Determine whether it is necessary to manually adjust the neighborhood radius range of each trusted label data.

[0264] If yes, then execute S909 below; otherwise, execute S910 below.

[0265] S909. Measure the characteristic distance between target points using a distance measurement tool, and set the neighborhood radius of the corresponding trusted label data based on the measured distance.

[0266] S910. Assign the labels of the trusted labeled data to the data within their respective extension ranges to complete the label extension of the dataset.

[0267] Secondly, in order to implement the above data processing method, a data processing apparatus according to an embodiment of this application is described below with a schematic diagram of the structure of the data processing apparatus.

[0268] like Figure 12 As shown, the data processing device 120 includes: a first determining unit 1201, a second determining unit 1202, a third determining unit 1203, and a processing unit 1204. Wherein:

[0269] The first determining unit 1201 is used to determine a first scatter point on a scatter plot; the scatter plot includes scatter points of a first type and scatter points of a second type, one scatter point of the first type represents one original labeled data; one scatter point of the second type represents one unlabeled data; the first scatter point is any scatter point of the first type; the first scatter point represents the first data;

[0270] The second determining unit 1202 is used to determine the neighborhood radius range of the first scatter point based on the scatter plot; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate.

[0271] The third determining unit 1203 is used to determine the extended region of the first scatter point within the neighborhood radius;

[0272] The processing unit 1204 is configured to configure the label of the second type of scatter points in the extended area as the first label of the first scatter point; the first label is used to characterize the data type to which the first data belongs.

[0273] In some embodiments, if the scatter plot includes a second scatter point, the label of the second scatter point is the first label, and the second scatter point is the first scatter point.

[0274] The second determining unit 1202 is specifically used to: determine a third scatter point on the scatter plot; the third scatter point is a scatter point of the second type, and the first distance between the third scatter point and the second scatter point is less than the distance between the second scatter point and other scatter points on the scatter plot besides the third scatter point;

[0275] The first distance is determined as the starting value of the neighborhood radius range of the second scatter point;

[0276] L fourth scatter points are determined on the scatter plot; the fourth scatter points are located inside a circle centered on the second scatter point and with a preset value as the radius; L is an integer greater than 1;

[0277] Among the L fourth scattered points, determine the second distance between every two of the fourth scattered points;

[0278] The maximum second distance value among the second distances between any two fourth scattered points is determined as the endpoint value of the neighborhood radius range of the second scattered point.

[0279] In some embodiments, if the scatter plot includes N second scatter points, where N is an integer greater than 1, the label of the second scatter point is the first label, and the second scatter point includes the first scatter point; the second determining unit 1202 is specifically used for:

[0280] For each of the N second scattered points, determine the initial value of the neighborhood radius of each second scattered point to obtain N initial values;

[0281] The smallest starting value among the N starting values ​​is determined as the starting value of the neighborhood radius range of the first scatter point;

[0282] For each of the N second scattered points, determine the endpoint value of the neighborhood radius of each second scattered point to obtain N endpoint values;

[0283] The minimum endpoint value among the N endpoint values ​​is determined as the endpoint value of the neighborhood radius range of the first scatter point.

[0284] In some embodiments, the third determining unit 1203 is configured to: determine the neighborhood radius of the i-th round; the neighborhood radius of the i-th round is within the range of the neighborhood radius; and i is an integer greater than or equal to 1;

[0285] Centered on the first scatter point, and with the neighborhood radius of the i-th round as the expansion radius, the expansion is carried out outward based on the first step length to obtain the expansion result of the i-th round; the expansion result of the i-th round is used to indicate that the expansion radius of the first scatter point in the i-th round is the neighborhood radius of the i-th round;

[0286] Based on the expansion results of the i-th round, determine the expansion radius of the first scatter point;

[0287] The extended region of the first scattered point is determined to be a circular region with the first scattered point as the center and the extended radius as the radius.

[0288] In some embodiments, the third determining unit 1203 is further configured to:

[0289] If the first expansion region of the i-th round includes a fifth scatter point, the label of the fifth scatter point is a second label, and the fifth scatter point is a scatter point of the first type; then cancel the expansion result of the i-th round, reduce the neighborhood radius of the (i+1)-th round, and determine the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)-th round.

[0290] If the first expansion region does not include the fifth scatter point, then the neighborhood radius of the (i+1)th round is increased, and the (i+1)th round is taken as the new i-th round, and the expansion continues until the first expansion region of the new i-th round includes the fifth scatter point.

[0291] In some embodiments, the third determining unit 1203 is further configured to:

[0292] If the first expansion region corresponding to the second data occupied in the i-th round intersects with the second expansion region corresponding to the second data occupied by the sixth scatter point, and the sixth scatter point is a scatter point of the first type, and the label of the sixth scatter point is different from that of the first scatter point; then the expansion radius of the first scatter point is determined based on the size relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data.

[0293] If the first expansion region and the second expansion region do not intersect, the neighborhood radius of the (i+1)th round is increased, and the (i+1)th round is taken as the new i-th round. Expansion continues until the first expansion region corresponding to the second data in the new i-th round intersects with the second expansion region when the sixth scatter point occupies the second data.

[0294] In some embodiments, the third determining unit 1203 is further configured to:

[0295] If the neighborhood radius of the i-th round is greater than or equal to the second neighborhood radius, then the expansion result of the i-th round is cancelled, and the neighborhood radius of the (i+1)-th round is reduced. The expansion radius of the first scatter point is determined based on the reduced neighborhood radius of the (i+1)-th round.

[0296] If the neighborhood radius of the i-th round is smaller than the second neighborhood radius, then the neighborhood radius of the (i+1)-th round is increased, and the (i+1)-th round is taken as the new i-th round, and the expansion continues.

[0297] In some embodiments, the third determining unit 1203 is further configured to:

[0298] If the neighborhood radius of the (i+1)th round is less than or equal to the neighborhood radius of the previous round, then the neighborhood radius of the previous round is determined as the expansion radius of the first scatter point.

[0299] If the neighborhood radius of round i+1 is greater than the neighborhood radius of the previous round, then the neighborhood radius of round i+2 is increased, and round i+2 is taken as the new round i+1, and the expansion continues until the neighborhood radius of the new round i+1 is less than or equal to the neighborhood radius of the previous round.

[0300] In some embodiments, the data processing apparatus 120 further includes an adjustment unit, the adjustment unit being configured to:

[0301] Calculate the third distance between the first scattered point and M seventh scattered points to obtain M third distances; the extended regions of the seventh scattered points and the first scattered points satisfy a specified relationship; M is an integer greater than 0;

[0302] If H of the M third distances satisfy an unspecified condition with the neighborhood radius of the extended region, a first operation is received, and the extended region of the first scatter point is adjusted based on the first operation; where H is less than or equal to M.

[0303] In some embodiments, the data processing apparatus 120 further includes a display unit, the display unit being used for:

[0304] Within the extended area corresponding to the first scatter point on the scatter plot, the first scatter point and the second type of scatter points are displayed in the same form.

[0305] It should be noted that the data processing device provided in this application embodiment includes all the units included, which can be implemented by a processor in an electronic device; of course, it can also be implemented by specific logic circuits; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field-programmable gate array (FPGA), etc.

[0306] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0307] It should be noted that, in the embodiments of this application, if the above-described data processing method is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0308] Thirdly, to implement the above data processing method, embodiments of this application provide an electronic device, including a memory and a processor. The memory stores a computer program that can run on the processor, and when the processor executes the program, it implements the steps in the data processing method provided in the above embodiments.

[0309] The following is combined Figure 13 The electronic device 130 shown is illustrated with a structural diagram of the electronic device.

[0310] In one example, electronic device 130 can be the aforementioned electronic device. For example... Figure 13 As shown, the electronic device 130 includes: a processor 1301, at least one communication bus 1302, a user interface 1303, at least one external communication interface 1304, and a memory 1305. The communication bus 1302 is configured to enable communication between these components. The user interface 1303 may include a display screen, and the external communication interface 1304 may include standard wired and wireless interfaces.

[0311] The memory 1305 is configured to store instructions and applications executable by the processor 1301, and can also cache data to be processed or already processed by the processor 1301 and various modules in the electronic device (e.g., image data, audio data, voice communication data and video communication data), which can be implemented by flash memory or random access memory (RAM).

[0312] Fourthly, embodiments of this application provide a storage medium, namely a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, implements the steps in the data processing method provided in the above embodiments.

[0313] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0314] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0315] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0316] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0317] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0318] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0319] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0320] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0321] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A data processing method, the method being applied to a data processing device, the method comprising: Obtain the data to be labeled and the labeled data; The data to be labeled includes image data, text data, or audio data; Based on the data to be labeled and the labeled data, a scatter plot is drawn based on the similarity between the data; the scatter plot includes a first type of scatter points and a second type of scatter points, where one first type of scatter point represents one original labeled data point; and one second type of scatter point represents one unlabeled data point. On the scatter plot, a first scatter point is determined, which is any scatter point of a first type and represents the first data. Based on the scatter plot, the neighborhood radius range of the first scatter point is determined; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate to. Within the radius of the neighborhood, determine the extended region of the first scatter point; Configure the label of the second type of scatter points within the extended region as the first label of the first scatter points; The first label is used to characterize the data type to which the first data belongs.

2. The method of claim 1, wherein if the scatter plot includes a second scatter point, the second scatter point having the same label as the first label, the second scatter point is the first scatter point. Determining the neighborhood radius range of the first scatter point based on the scatter plot includes: A third scatter point is determined on the scatter plot; the third scatter point is a scatter point of the second type, and the first distance between the third scatter point and the second scatter point is less than the distance between the second scatter point and other scatter points on the scatter plot besides the third scatter point. The first distance is determined as the starting value of the neighborhood radius range of the second scatter point; L fourth scatter points are determined on the scatter plot; the fourth scatter points are located inside a circle centered on the second scatter point and with a preset value as the radius; L is an integer greater than 1; Among the L fourth scattered points, determine the second distance between every two of the fourth scattered points; The maximum second distance value among the second distances between any two fourth scattered points is determined as the endpoint value of the neighborhood radius range of the second scattered point.

3. The method of claim 1, wherein if the scatter plot includes N second scatter points, the N being an integer greater than 1, the second scatter points having the first label, the second scatter points including the first scatter point. Determining the neighborhood radius range of the first scatter point based on the scatter plot includes: For each of the N second scattered points, determine the initial value of the neighborhood radius of each second scattered point to obtain N initial values; The smallest starting value among the N starting values ​​is determined as the starting value of the neighborhood radius range of the first scatter point; For each of the N second scattered points, determine the endpoint value of the neighborhood radius of each second scattered point to obtain N endpoint values; The minimum endpoint value among the N endpoint values ​​is determined as the endpoint value of the neighborhood radius range of the first scatter point.

4. The method according to claim 1, wherein determining the extended region of the first scatter point within the neighborhood radius includes: Determine the neighborhood radius in the i-th round; The neighborhood radius of the i-th round is within the range of the neighborhood radius; The i is an integer greater than or equal to 1; Centered on the first scattered point, and with the neighborhood radius of the i-th round as the expansion radius, expand outward based on the first step length to obtain the expansion result of the i-th round; The expansion result of the i-th round is used to indicate that the expansion radius of the first scatter point in the i-th round is the neighborhood radius of the i-th round; Based on the expansion results of the i-th round, determine the expansion radius of the first scatter point; The extended region of the first scattered point is determined to be a circular region with the first scattered point as the center and the extended radius as the radius.

5. The method according to claim 4, wherein determining the expansion radius of the first scatter point based on the expansion result of the i-th round comprises: If the first expansion region of the i-th round includes a fifth scatter point, the label of the fifth scatter point is a second label, and the fifth scatter point is a scatter point of the first type; then cancel the expansion result of the i-th round, reduce the neighborhood radius of the (i+1)-th round, and determine the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)-th round. If the first expansion region does not include the fifth scatter point, then the neighborhood radius of the (i+1)th round is increased, and the (i+1)th round is taken as the new i-th round, and the expansion continues until the first expansion region of the new i-th round includes the fifth scatter point.

6. The method according to claim 4, wherein determining the expansion radius of the first scatter point based on the expansion result of the i-th round comprises: If the first expansion region corresponding to the second data occupied in the i-th round intersects with the second expansion region corresponding to the second data occupied by the sixth scatter point, and the sixth scatter point is a scatter point of the first type, and the label of the sixth scatter point is different from that of the first scatter point; then the expansion radius of the first scatter point is determined based on the size relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data. If the first expansion region and the second expansion region do not intersect, the neighborhood radius of the (i+1)th round is increased, and the (i+1)th round is taken as the new i-th round. Expansion continues until the first expansion region corresponding to the second data in the new i-th round intersects with the second expansion region when the sixth scatter point occupies the second data.

7. The method according to claim 6, wherein determining the expansion radius of the first scatter point based on the relationship between the neighborhood radius of the i-th round and the second neighborhood radius when the sixth scatter point occupies the second data includes: If the neighborhood radius of the i-th round is greater than or equal to the second neighborhood radius, then the expansion result of the i-th round is cancelled, and the neighborhood radius of the (i+1)-th round is reduced. The expansion radius of the first scatter point is determined based on the reduced neighborhood radius of the (i+1)-th round. If the neighborhood radius of the i-th round is smaller than the second neighborhood radius, then the neighborhood radius of the (i+1)-th round is increased, and the (i+1)-th round is taken as the new i-th round, and the expansion continues.

8. The method according to claim 5 or 7, wherein determining the expansion radius of the first scatter point based on the reduced neighborhood radius of the (i+1)th round comprises: If the neighborhood radius of the (i+1)th round is less than or equal to the neighborhood radius of the previous round, then the neighborhood radius of the previous round is determined as the expansion radius of the first scatter point. If the neighborhood radius of round i+1 is greater than the neighborhood radius of the previous round, then the neighborhood radius of round i+2 is increased, and round i+2 is taken as the new round i+1, and the expansion continues until the neighborhood radius of the new round i+1 is less than or equal to the neighborhood radius of the previous round.

9. The method according to claim 1, The method further includes: Calculate the third distance between the first scattered point and the M seventh scattered points to obtain the M third distances; The extended regions of the seventh scatter point and the first scatter point satisfy a specified relationship; M is an integer greater than 0; the specified relationship is used to characterize the actual extended region of the seventh scatter point that did not originally belong to the first scatter point; If H of the M third distances satisfy a non-specified condition with the neighborhood radius of the extended region, a first operation is received, and the extended region of the first scatter point is adjusted based on the first operation; H is less than or equal to M; the non-specified condition corresponds to a specified relationship and is used to characterize that the extended region of the first scatter point includes the seventh scatter point.

10. A data processing apparatus, the apparatus being deployed in a data processing device, the apparatus comprising: The acquisition unit is used to acquire data to be labeled and already labeled data. The data to be labeled includes image data, text data, or audio data; A plotting unit is used to plot a scatter plot based on the similarity between the data to be labeled and the labeled data; the scatter plot includes a first type of scatter plot and a second type of scatter plot, where a first type of scatter plot represents an original labeled data point and a second type of scatter plot represents an unlabeled data point; The first determining unit is configured to determine a first scatter point on the scatter plot, wherein the first scatter point is any scatter point of a first type, representing the first data; The second determining unit is used to determine the neighborhood radius range of the first scatter point based on the scatter plot; the neighborhood radius range is used to characterize the range of values ​​of the neighborhood radius that the first scatter point can propagate. The third determining unit is used to determine the extended region of the first scatter point within the neighborhood radius; Processing unit, configured to configure the labels of the second type of scatter points in the extended area as the first labels of the first scatter points; The first label is used to characterize the data type to which the first data belongs.

Images