Sample determination method, model training method, device and equipment

Abstract

The present disclosure provides a sample determination method, a model training method, an apparatus and a device, which can be applied to the fields of artificial intelligence technology and financial technology. The method comprises: determining, for each positive sample in a positive sample set, a first sample set from a benchmark sample set, wherein the similarity between each benchmark sample in the first sample set and the positive sample satisfies a similarity threshold; determining, for each negative sample in a negative sample set, a first sample subset from the benchmark sample set, wherein the similarity between each benchmark sample in the first sample subset and the negative sample satisfies the similarity threshold; determining, based on the first sample set, a second sample set from a plurality of first sample subsets; and determining the positive sample as a target positive sample in a case where the first sample set and the second sample set satisfy a predetermined condition.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of artificial intelligence and the field of financial technology, and in particular to a sample determination method, a model training method, an apparatus, a device, a medium and a program product. BACKGROUND

[0002] In the prior art, there are a large number of unlabeled data in the financial industry. In the training process of the model, the labeled data is mainly used for training. Since the labeled data is obtained by manually labeling the unlabeled data.

[0003] In the process of implementing the present disclosure, the inventors have found that in the related art, the target data is determined from the complex labeled data in an artificial manner, which wastes human resources and is low in efficiency. SUMMARY

[0004] In view of the above problems, the present disclosure provides a sample determination method, a model training method, an apparatus, a device, a medium and a program product.

[0005] According to a first aspect of the present disclosure, a sample determination method is provided, comprising: determining a first sample set from a reference sample set for each positive sample in a positive sample set, wherein the similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold; determining a first sample subset from the reference sample set for each negative sample in a negative sample set, wherein the similarity between each reference sample in the first sample subset and the negative sample satisfies the similarity threshold; determining a second sample set from a plurality of first sample subsets based on the first sample set; and determining the positive sample as a target positive sample in a case where the first sample set and the second sample set satisfy a predetermined condition.

[0006] According to an embodiment of the present disclosure, before determining the first sample set from the reference sample set, the method further comprises: determining the positive sample set, the negative sample set and the reference sample set from a sample database in response to a sample determination instruction from an electronic device; calling a sample transmission interface to obtain the positive sample set, the negative sample set and the reference sample set from the sample database.

[0007] According to an embodiment of the present disclosure, the positive sample comprises positive sample data, the positive sample data comprising: asset data, object type data, storage duration data, transaction data; the negative sample comprises negative sample data, the negative sample data comprising: asset data, object type data, storage duration data, transaction data; the reference sample comprises reference sample data, the reference sample data comprising: asset data, object type data, storage duration data, transaction data.

[0008] According to an embodiment of the present disclosure, before determining the positive sample as the target positive sample in a case where it is determined that the first sample set and the second sample set satisfy a predetermined condition, the method further includes: determining a first sample quantity of the first sample set; determining a second sample quantity of the second sample set; and determining that the first sample set and the second sample set satisfy the predetermined condition in a case where it is determined that a ratio between the first sample quantity and the second sample quantity is greater than a predetermined threshold value.

[0009] According to an embodiment of the present disclosure, the sample determination method further includes: determining, for each negative sample in the negative sample set, a third sample set from the reference sample set, wherein a similarity between each reference sample in the third sample set and the negative sample satisfies a similarity threshold value; determining, for each positive sample in the positive sample set, a second sample subset from the reference sample set, wherein a similarity between each reference sample in the second sample subset and the positive sample satisfies the similarity threshold value; determining, based on the third sample set, a fourth sample set from the plurality of second sample subsets; and determining the negative sample as a target negative sample in a case where it is determined that the third sample set and the fourth sample set satisfy a predetermined condition.

[0010] According to an embodiment of the present disclosure, before determining the negative sample as the target negative sample in a case where it is determined that the third sample set and the fourth sample set satisfy a predetermined condition, the sample determination method further includes: determining a third sample quantity of the third sample set; determining a fourth sample quantity of the fourth sample set; and determining that the third sample set and the fourth sample set satisfy the predetermined condition in a case where it is determined that a ratio between the third sample quantity and the fourth sample quantity is greater than a predetermined threshold value.

[0011] According to an embodiment of the present disclosure, the positive sample set includes a plurality of positive sample data, and the reference sample includes a plurality of reference sample data; before determining the first sample set from the reference sample set, the sample determination method further includes: determining, from the plurality of positive sample data, continuous target positive sample data of a continuous type; determining, from the plurality of reference sample data, continuous target first reference sample data matched with the continuous target positive sample data; determining, from the plurality of positive sample data, discrete target positive sample data of a discrete type; determining, from the plurality of reference sample data, discrete target first reference sample data matched with the discrete target positive sample data; determining a first continuous sample similarity based on the continuous target positive sample data and the continuous target first reference sample data; determining a first discrete sample similarity based on the discrete target positive sample data and the discrete target first reference sample data; and determining a first target similarity between the positive sample and the reference sample based on the first continuous sample similarity and the first discrete sample similarity.

[0012] According to an embodiment of the present disclosure, the first continuous sample similarity is determined based on the continuous target positive sample data and the continuous target first reference sample data, including: determining a first feature vector of the continuous target positive sample data; determining a second feature vector of the continuous target first reference sample data; and determining the first continuous sample similarity based on the first feature vector and the second feature vector.

[0013] According to an embodiment of the present disclosure, the first discrete sample similarity is determined based on the discrete target positive sample data and the discrete target first reference sample data, including: in a case where it is determined that the discrete target positive sample data and the discrete target first reference sample data are the same, determining the first discrete sample similarity based on the first continuous sample similarity; and in a case where it is determined that the discrete target positive sample data and the discrete target first reference sample data are not the same, determining the first discrete sample similarity as a first predetermined similarity.

[0014] According to an embodiment of the present disclosure, the negative sample set includes a plurality of negative sample data, and the reference sample set includes a plurality of reference sample data; before the first sample subset is determined from the reference sample set, the sample determination method further includes: determining continuous target negative sample data of a continuous type from the plurality of negative sample data; determining continuous target second reference sample data matched with the continuous target negative sample data from the plurality of reference sample data; determining discrete target negative sample data of a discrete type from the plurality of negative sample data; determining discrete target second reference sample data matched with the discrete target negative sample data from the plurality of reference sample data; determining a second continuous sample similarity based on the continuous target negative sample data and the continuous target second reference sample data; determining a second discrete sample similarity based on the discrete target negative sample data and the discrete target second reference sample data; and determining a second target similarity between the negative sample and the reference sample based on the second continuous sample similarity and the second discrete sample similarity.

[0015] According to an embodiment of the present disclosure, the second continuous sample similarity is determined based on the continuous target negative sample data and the continuous target second reference sample data, including: determining a third feature vector of the continuous target negative sample data; determining a fourth feature vector of the continuous target second reference sample data; and determining the second continuous sample similarity based on the third feature vector and the fourth feature vector.

[0016] According to an embodiment of the present disclosure, the second discrete sample similarity is determined based on the discrete target negative sample data and the discrete target second reference sample data, including: in a case where it is determined that the discrete target negative sample data and the discrete target second reference sample data are the same, determining the second discrete sample similarity based on the second continuous sample similarity; and in a case where it is determined that the discrete target negative sample data and the discrete target second reference sample data are not the same, determining the second discrete sample similarity as a second predetermined similarity.

[0017] The second aspect of the present disclosure provides a model training method, comprising: determining a target positive sample from the fifth sample set according to the above-mentioned sample determination method; removing the target positive sample from the fifth sample set to obtain a sixth sample set; and training an initial model based on the sixth sample set to obtain a target model.

[0018] The third aspect of the present disclosure provides a sample determination apparatus, comprising: a first determination module configured to determine, for each positive sample in a positive sample set, a first sample set from a reference sample set, wherein the similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold; a second determination module configured to determine, for each negative sample in a negative sample set, a first sample subset from the reference sample set, wherein the similarity between each reference sample in the first sample subset and the negative sample satisfies the similarity threshold; a third determination module configured to determine, based on the first sample set, a second sample set from a plurality of first sample subsets; and a fourth determination module configured to determine that the positive sample is a target positive sample in a case where the first sample set and the second sample set satisfy a predetermined condition.

[0019] The fourth aspect of the present disclosure provides a model training apparatus, comprising: a fifth determination module configured to determine a target positive sample from a fifth sample set according to the above-mentioned sample determination method; a removal module configured to remove the target positive sample from the fifth sample set to obtain a sixth sample set; and a training module configured to train an initial model based on the sixth sample set to obtain a target model.

[0020] The fifth aspect of the present disclosure provides an electronic device, comprising: one or more processors; a memory configured to store one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the above-mentioned method.

[0021] The sixth aspect of the present disclosure further provides a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the above-mentioned method.

[0022] The seventh aspect of the present disclosure further provides a computer program product comprising a computer program which, when executed by a processor, implements the above-mentioned method.

[0023] According to the sample determination method, the model training method, the apparatus, the device, the medium and the program product provided by the present disclosure, the first sample set determined by the positive sample and the plurality of first sample subsets determined by the negative sample set are used to determine the second sample set from the plurality of first sample subsets based on the first sample set, and then the positive sample is determined as the target positive sample according to the first sample set and the second sample set, which fully utilizes the computing power, achieves the purpose of automatically determining the target positive sample, saves resources, and improves the efficiency of determining the target positive sample. BRIEF DESCRIPTION OF DRAWINGS

[0024] The above and other objects, features and advantages of the present disclosure will become more apparent from the following description when taken in conjunction with the accompanying drawings, in which:

[0025] Figure 1 A scenario diagram of application of a sample determination method or a model training method according to embodiments of the present disclosure is schematically shown;

[0026] Figure 2 A flowchart of a sample determination method according to embodiments of the present disclosure is schematically shown;

[0027] Figure 3 A schematic diagram of a method of determining a first sample set according to embodiments of the present disclosure is schematically shown;

[0028] Figure 4 A schematic diagram of a method of determining a first sample subset according to embodiments of the present disclosure is schematically shown;

[0029] Figure 5 A schematic diagram of a method of determining a first target similarity according to embodiments of the present disclosure is schematically shown;

[0030] Figure 6 A schematic diagram of a method of determining a second target similarity according to embodiments of the present disclosure is schematically shown;

[0031] Figure 7 A flowchart of a model training method according to embodiments of the present disclosure is schematically shown;

[0032] Figure 8 A structural block diagram of a sample determination apparatus according to embodiments of the present disclosure is schematically shown;

[0033] Figure 9 A structural block diagram of a model training apparatus according to embodiments of the present disclosure is schematically shown;

[0034] Figure 10 A block diagram of an electronic device suitable for implementing a sample determination method or a model training method according to embodiments of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0035] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It is to be understood, however, that the description is merely exemplary of the present disclosure, and is not intended to limit the scope of the present disclosure. In the following detailed description of embodiments of the present disclosure, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. However, it would be apparent to one skilled in the art that one or more embodiments of the present disclosure can be practiced without these specific details. In other instances, well-known structures and functions have not been described in detail in order to avoid obscuring aspects of the present disclosure.

[0036] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used herein, the terms "comprises", "comprising", "includes", "including" and the like are specifically intended to be open-ended and to mean that other features, steps, operations, and / or components can be added.

[0037] All terms used herein, including technical and scientific terms, have the meanings commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein are defined as having meanings that are consistent with the context of the specification in which the terms are used and should not be explained in an idealized or overly formal manner.

[0038] In the case of using expressions similar to "at least one of A, B, and C, etc.", it should generally be interpreted to include at least one of each item enumerated. For example, "a system having at least one of A, B, and C" should be interpreted to include a system having A alone, a system having B alone, a system having C alone, a system having both A and B, a system having both A and C, a system having both B and C, and / or a system having A, B, and C, etc.

[0039] In the technical solutions of the disclosure, the collection, storage, use, processing, transmission, provision, disclosure, and application of data (such as including but not limited to user personal information) involved in the technical solutions comply with the relevant legal regulations, necessary security measures are taken, and the public order and good customs are not violated.

[0040] The training effect of a machine learning model is affected by factors such as data, algorithms, and computing power, among which data is a more critical part. In the financial business application scenario, the labeled data is usually used more widely and has better effect.

[0041] In the process of implementing the concept of the disclosure, the inventors found that in the related art, the target labeled data is determined from the complex labeled data in an artificial manner, which wastes human resources and is low in efficiency.

[0042] Therefore, an embodiment of the disclosure provides a sample determination method, including: determining, for each positive sample in a positive sample set, a first sample set from a benchmark sample set, wherein a similarity between each benchmark sample in the first sample set and the positive sample satisfies a similarity threshold; determining, for each negative sample in a negative sample set, a first sample subset from the benchmark sample set, wherein a similarity between each benchmark sample in the first sample subset and the negative sample satisfies the similarity threshold; determining, based on the first sample set, a second sample set from a plurality of first sample subsets; and determining the positive sample as a target positive sample in a case where the first sample set and the second sample set satisfy a predetermined condition.

[0043] According to an embodiment of the present disclosure, the electronic device can first call a sample transmission interface from a sample database to obtain a positive sample set, a negative sample set and a benchmark sample set. The positive sample set, the negative sample set and the benchmark sample set can each include asset data, object type data, storage duration data and transaction data in the financial field. The storage duration data may, for example, include storage duration data of storage amount data in the asset data. The electronic device then processes the positive sample, the negative sample and the benchmark sample constructed according to the data in the financial field to determine the positive sample as a target positive sample. Since the target positive sample is determined automatically by the electronic device, the computing power of the electronic device is fully utilized to efficiently process a large amount of complex asset data, object type data, storage duration data and transaction data in the financial field, thereby solving the problem of low efficiency of the manual determination method.

[0044] Figure 1 An application scenario diagram of a sample determination method or a model training method according to an embodiment of the present disclosure is schematically shown.

[0045] As shown in Figure 1 The application scenario 100 according to the embodiment can include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0046] A user can use at least one of the first terminal device 101, the second terminal device 102 and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102 and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0047] The first terminal device 101, the second terminal device 102 and the third terminal device 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smartphones, tablet computers, laptop computers and desktop computers, etc.

[0048] The server 105 can be a server that provides various services, such as a background management server that provides support for a website browsed by a user using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (as an example only). The background management server can perform analysis and the like on received user requests and the like, and feed back a processing result (such as a web page, information, or data obtained or generated according to a user request, and the like) to the terminal device.

[0049] It should be noted that the sample determination method or the model training method provided by the embodiments of the present disclosure can generally be executed by the server 105. Accordingly, the sample determination apparatus or the model training apparatus provided by the embodiments of the present disclosure can generally be arranged in the server 105. The sample determination method or the model training method provided by the embodiments of the present disclosure can also be executed by a server or a server cluster different from the server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105. Accordingly, the sample determination apparatus or the model training apparatus provided by the embodiments of the present disclosure can also be arranged in a server or a server cluster different from the server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or the server 105.

[0050] It should be understood that Figure 1 The number of terminal devices, networks, and servers in the above-described scenario is merely illustrative. Any number of terminal devices, networks, and servers can be provided according to implementation needs.

[0051] The sample determination method according to the embodiments of the present disclosure will be described in detail below based on the scenario described above. Figure 1 Figures 2-7 The sample determination method according to the embodiments of the present disclosure will be described in detail below based on the scenario described above.

[0052] Figure 2 A flowchart of a sample determination method according to an embodiment of the present disclosure is schematically shown.

[0053] As shown in Figure 2 , the sample determination method of this embodiment includes operation S210 to operation S240.

[0054] In operation S210, for each positive sample in the positive sample set, a first sample set is determined from the reference sample set, wherein the similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold.

[0055] ​According to an embodiment of the present disclosure, the positive sample can be a sample to be determined as a target positive sample. The benchmark sample set can include a plurality of benchmark samples, and the benchmark sample set can be used to determine a benchmark sample having a similarity to the positive sample satisfying a similarity threshold, and obtain the first sample set according to the benchmark sample. The similarity threshold can be 70%, 80%, 90%, etc., and the present disclosure does not limit this.

[0056] For example, the positive sample set can include a positive sample A1; the benchmark sample set can include a benchmark sample B1, a benchmark sample B2, a benchmark sample B3, a benchmark sample B4, and a benchmark sample B5, wherein the benchmark samples having a similarity to the positive sample A1 satisfying a similarity threshold can be the benchmark sample B1 and the benchmark sample B2. Thus, the first sample set can include the benchmark sample B1 and the benchmark sample B2.

[0057] According to an embodiment of the present disclosure, the similarity between the benchmark sample and the positive sample can be determined according to the positive sample data included in each positive sample and the benchmark sample data included in the benchmark sample; the model can also be trained using the positive sample and the model can also be trained using the benchmark sample, and the similarity between the effect of the model trained by the positive sample and the effect of the model trained by the benchmark sample can be compared to determine the similarity between the positive sample and the benchmark sample.

[0058] In operation S220, for each negative sample in the negative sample set, a first sample subset is determined from the benchmark sample set, wherein the similarity between each benchmark sample in the first sample subset and the negative sample satisfies a similarity threshold.

[0059] According to an embodiment of the present disclosure, the negative sample can be used to determine that the positive sample is a target positive sample. For example, by determining the data similarity between the positive sample data included in the positive sample and the negative sample data included in the negative sample, in the case where the data similarity is greater than a predetermined threshold, it can be determined that the positive sample is a target positive sample.

[0060] The benchmark sample set can also be used to determine a benchmark sample having a similarity to the negative sample satisfying a similarity threshold, and obtain the first sample subset according to the benchmark sample.

[0061] For example, the negative sample set can include negative sample C1, negative sample C2, negative sample C3, and negative sample C4; the benchmark sample set can include benchmark sample B1, benchmark sample B2, benchmark sample B3, benchmark sample B4, benchmark sample B5, benchmark sample B6, benchmark sample B7, benchmark sample B8, and benchmark sample B9, wherein the benchmark samples similar to negative sample C1 and satisfying the similarity threshold can be benchmark sample B1 and benchmark sample B2, and thus the first sample subset D1 can include benchmark sample B1 and benchmark sample B2; the benchmark samples similar to negative sample C2 and satisfying the similarity threshold can be benchmark sample B2 and benchmark sample B3, and thus the first sample subset D2 can include benchmark sample B2 and benchmark sample B3; the benchmark samples similar to negative sample C3 and satisfying the similarity threshold can be benchmark sample B4 and benchmark sample B5, and thus the first sample subset D3 can include benchmark sample B4 and benchmark sample B5; the benchmark samples similar to negative sample C4 and satisfying the similarity threshold can be benchmark sample B6, and thus the first sample subset D4 can include benchmark sample B6.

[0062] According to an embodiment of the present disclosure, the similarity between the benchmark sample and the negative sample can be determined according to the negative sample data included in the negative sample and the benchmark sample data included in the benchmark sample; the similarity between the effect of the model trained by the negative sample and the effect of the model trained by the benchmark sample can be compared by training the model by using the negative sample and training the model by using the benchmark sample, so as to determine the similarity between the negative sample and the benchmark sample.

[0063] In operation S230, the second sample set is determined from the plurality of first sample subsets based on the first sample set.

[0064] According to an embodiment of the present disclosure, the second sample set can include the benchmark samples in the plurality of first sample subsets and satisfying the similarity threshold with the first sample set.

[0065] For example, the first sample subset D1 can include benchmark sample B1 and benchmark sample B2; the first sample subset D2 can include benchmark sample B2 and benchmark sample B3; the first sample subset D3 can include benchmark sample B4 and benchmark sample B5; and the first sample subset D4 can include benchmark sample B6. Thus, the plurality of first sample subsets can include benchmark sample B1, benchmark sample B2, benchmark sample B3, benchmark sample B4, benchmark sample B5, and benchmark sample B6, and the first sample set can include benchmark sample B1 and benchmark sample B2.

[0066] Thus, the second sample set can include the benchmark samples B1 and B2 in the first sample set and the plurality of first sample subsets and having a similarity of 100%.

[0067] According to an embodiment of the present disclosure, similarity between the reference sample included in the first sample set and the reference sample included in the plurality of first sample subsets can be determined, and the second sample subset can be obtained according to the reference samples in the plurality of first sample subsets that satisfy the similarity threshold.

[0068] In operation S240, in a case where it is determined that the first sample set and the second sample set satisfy the predetermined condition, the positive sample is determined as the target positive sample.

[0069] According to an embodiment of the present disclosure, the proportion of the number of the reference sample in the second sample set in the first sample set can be determined according to the number of samples in the second sample set and the number of samples in the first sample set. Thus, whether the sample data difference between the positive sample and the negative sample satisfies the requirement can be determined. If the proportion of the number is high, it is determined that the sample data difference between the positive sample and the negative sample does not satisfy the requirement. Thus, the positive sample can be determined as the target positive sample.

[0070] For example, the first sample set can include the reference sample B1 and the reference sample B2; and the second sample set can include the reference sample B1 and the reference sample B2. Thus, the proportion of the number of samples in the second sample set in the first sample set is 100%, and the positive sample can be determined as the target positive sample.

[0071] According to an embodiment of the present disclosure, based on the first sample set determined by the positive sample and the plurality of first sample subsets determined by the negative sample set, the second sample set is determined from the plurality of first sample subsets based on the first sample set, and then the positive sample is determined as the target positive sample according to the first sample set and the second sample set. The purpose of automatically determining the target positive sample is achieved, resources are saved, computing power is fully utilized, and the efficiency of determining the target positive sample is improved.

[0072] Figure 3 A schematic diagram of a method for determining a first sample set according to an embodiment of the present disclosure is schematically shown.

[0073] As Figure 3 shown, the determination that the target similarity satisfies the predetermined similarity threshold includes operations S310-S360.

[0074] In operation S310, the positive sample is obtained.

[0075] In operation S320, the reference sample is determined.

[0076] In operation S330, the similarity is determined.

[0077] In operation S340, it is determined whether the similarity satisfies the predetermined similarity threshold. If yes, operation S360 is performed; if no, operation S350 is performed.

[0078] In operation S350, the positive sample is stopped.

[0079] In operation S360, the reference sample is added to the first sample set.

[0080] Figure 4 A schematic diagram of a method for determining a first sample subset according to an embodiment of the present disclosure is shown schematically.

[0081] As Figure 4 The embodiment determines whether the target similarity meets a predetermined similarity threshold, including operations S410-S460.

[0082] In operation S410, the negative sample is obtained.

[0083] In operation S420, the reference sample is determined.

[0084] In operation S430, the similarity is determined.

[0085] In operation S440, it is determined whether the similarity meets a predetermined similarity threshold. If yes, operation S460 is performed; if no, operation S450 is performed.

[0086] In operation S450, the negative sample is stopped.

[0087] In operation S460, the reference sample is added to the first sample subset.

[0088] According to an embodiment of the present disclosure, the positive sample includes positive sample data, the positive sample data including asset data, object type data, storage duration data, and transaction data; the negative sample includes negative sample data, the negative sample data including asset data, object type data, storage duration data, and transaction data; and the reference sample includes reference sample data, the reference sample data including asset data, object type data, storage duration data, and transaction data.

[0089] According to an embodiment of the present disclosure, by using the sample determination method of the present disclosure, a large amount of complex asset data, object type data, storage duration data, and transaction data in the financial field are processed to determine the required target positive sample, thereby improving the data processing efficiency and meeting the demand for efficiency.

[0090] According to an embodiment of the present disclosure, before determining the first sample set from the reference sample set, the method further includes: in response to a sample determination instruction from the electronic device, determining the positive sample set, the negative sample set, and the reference sample set from the sample database; and calling a sample transmission interface to obtain the positive sample set, the negative sample set, and the reference sample set from the sample database.

[0091] According to an embodiment of the present disclosure, the sample determination instruction can be from other electronic devices or pre-stored in an instruction database of the electronic device.

[0092] According to an embodiment of the present disclosure, the sample transmission interface can be a data interface pre-set on the electronic device, for example, a serial communication interface or the like.

[0093] According to an embodiment of the present disclosure, the sample database can be pre-stored on the electronic device or stored on other electronic devices, and the positive sample set, the negative sample set and the benchmark sample set constructed by the data in the financial field can be obtained from the other electronic devices through the sample transmission interface.

[0094] According to an embodiment of the present disclosure, the positive sample set, the negative sample set and the benchmark sample set are obtained from the sample database by calling the sample transmission interface according to the sample determination instruction by the electronic device, so that the efficient acquisition of a large number of complex samples is realized.

[0095] According to an embodiment of the present disclosure, before determining the positive sample as the target positive sample in the case where the first sample set and the second sample set satisfy the predetermined condition, the sample determination method further comprises: determining a first sample quantity of the first sample set; determining a second sample quantity of the second sample set; and determining that the first sample set and the second sample set satisfy the predetermined condition in the case where a ratio between the first sample quantity and the second sample quantity is greater than a predetermined threshold.

[0096] According to an embodiment of the present disclosure, the first sample quantity can be a sample quantity of all samples included in the first sample set, and the second sample quantity can be a sample quantity of all samples included in the second sample set.

[0097] The ratio between the first sample quantity and the second sample quantity, for example, in the case where the first sample set includes 5 benchmark samples and the second sample set includes 3 benchmark samples, the ratio between the first sample quantity and the second sample quantity can be 3 / 5, that is, 60%.

[0098] In the case where the predetermined threshold is 60%, it can be determined that the first sample set and the second sample set satisfy the predetermined condition, so that the positive sample can be determined as the target positive sample. It should be noted that the predetermined threshold can be 70%, 80%, 90%, 100% or the like, and the present disclosure does not limit the predetermined threshold.

[0099] According to an embodiment of the present disclosure, since the target positive sample is determined according to the ratio between the first sample quantity and the second sample quantity by the first sample quantity and the second sample quantity, the accuracy of determining the target positive sample is improved.

[0100] According to an embodiment of the present disclosure, the sample determination method further includes: determining, for each negative sample in the negative sample set, a third sample set from the reference sample set, wherein a similarity between each reference sample in the third sample set and the negative sample satisfies a similarity threshold; determining, for each positive sample in the positive sample set, a second sample subset from the reference sample set, wherein a similarity between each reference sample in the second sample subset and the positive sample satisfies the similarity threshold; determining, based on the third sample set, a fourth sample set from the plurality of second sample subsets; and determining the negative sample as the target negative sample in a case where the third sample set and the fourth sample set satisfy a predetermined condition.

[0101] According to an embodiment of the present disclosure, the reference sample set can also be used to determine a reference sample whose similarity with the negative sample satisfies the similarity threshold, and obtain the third sample set according to the reference sample.

[0102] For example, the negative sample set can include a negative sample C1; the reference sample set can include a reference sample B1, a reference sample B2, a reference sample B3, a reference sample B4, a reference sample B5, a reference sample B6, a reference sample B7, a reference sample B8, and a reference sample B9, wherein the reference samples whose similarity with the negative sample C1 satisfies the similarity threshold can be the reference sample B1 and the reference sample B2. Thus, the third sample set can include the reference sample B1 and the reference sample B2.

[0103] According to an embodiment of the present disclosure, the positive sample can also be used to determine the negative sample as the target negative sample. For example, according to a data similarity between the positive sample data included in the positive sample and the negative sample data included in the negative sample, it can be determined whether the negative sample is the target negative sample. In a case where the data similarity is greater than a predetermined threshold, the negative sample can be determined as the target negative sample.

[0104] The reference sample set can also be used to determine a reference sample whose similarity with the positive sample satisfies the similarity threshold, and obtain the second sample subset according to the reference sample.

[0105] For example, the positive sample set can include positive sample A1, positive sample A2, positive sample A3, and positive sample A4; the reference sample set can include reference sample B1, reference sample B2, reference sample B3, reference sample B4, reference sample B5, reference sample B6, reference sample B7, reference sample B8, and reference sample B9, wherein the reference samples similar to positive sample A1 and satisfying the similarity threshold can be reference sample B1 and reference sample B2, and thus reference sample B1 and reference sample B2 can be included in the second sample subset E1; the reference samples similar to positive sample A2 and satisfying the similarity threshold can be reference sample B2 and reference sample B3, and thus reference sample B2 and reference sample B3 can be included in the second sample subset E2; the reference samples similar to positive sample A3 and satisfying the similarity threshold can be reference sample B4 and reference sample B5, and thus reference sample B4 and reference sample B5 can be included in the third sample subset E3; the reference samples similar to positive sample A4 and satisfying the similarity threshold can be reference sample B6, and thus reference sample B6 can be included in the second sample subset E4.

[0106] According to an embodiment of the present disclosure, the fourth sample set can include reference samples in the plurality of second sample subsets and satisfying the similarity threshold with the third sample set.

[0107] According to an embodiment of the present disclosure, the similarity between the reference samples included in the third sample set and the reference samples included in the plurality of second sample subsets can be determined, and the fourth sample set can be obtained according to the reference samples in the plurality of second sample subsets and satisfying the similarity threshold.

[0108] For example, reference sample B1 and reference sample B2 can be included in the second sample subset E1; reference sample B2 and reference sample B3 can be included in the second sample subset E2; reference sample B4 and reference sample B5 can be included in the second sample subset E3; and reference sample B6 can be included in the second sample subset E4. Thus, the plurality of second sample subsets can include reference sample B1, reference sample B2, reference sample B3, reference sample B4, reference sample B5, and reference sample B6, and the third sample set can include reference sample B1 and reference sample B2.

[0109] Thus, the fourth sample set can include reference sample B1 and reference sample B2 in the third sample set and the plurality of second sample subsets and having a similarity of 100%.

[0110] According to an embodiment of the present disclosure, the proportion of the number of the reference samples in the third sample set in the fourth sample set can be determined according to the number of samples in the third sample set and the number of samples in the fourth sample set. In this way, whether the sample data difference between the positive samples and the negative samples meets the requirement can be determined. If the proportion of the number is high, it is determined that the sample data difference between the positive samples and the negative samples does not meet the requirement, and thus the negative sample can be determined as the target negative sample.

[0111] For example, the third sample set can include reference sample B1 and reference sample B2; the fourth sample set can include reference sample B1 and reference sample B2; and thus the proportion of the number of the samples in the third sample set in the fourth sample set is 100%, and the negative sample can be determined as the target negative sample.

[0112] According to an embodiment of the present disclosure, based on the third sample set determined by the negative sample and the plurality of second sample subsets determined by the positive sample set, the fourth sample set is determined from the plurality of second sample subsets based on the third sample set, and then the negative sample is determined as the target negative sample according to the third sample set and the fourth sample set. The computing power is fully utilized, the purpose of automatically determining the target negative sample is achieved, the resources are saved, and the efficiency of determining the target negative sample is improved.

[0113] According to an embodiment of the present disclosure, before the negative sample is determined as the target negative sample in the case where it is determined that the third sample set and the fourth sample set meet the predetermined condition, the sample determination method further includes: determining a third sample number of the third sample set; determining a fourth sample number of the fourth sample set; and in the case where it is determined that the ratio between the third sample number and the fourth sample number is greater than a predetermined threshold, determining that the third sample set and the fourth sample set meet the predetermined condition.

[0114] According to an embodiment of the present disclosure, the third sample number can be the number of all samples included in the third sample set; and the fourth sample number can be the number of all samples included in the fourth sample set.

[0115] The ratio between the third sample number and the second sample number, for example, in the case where the third sample set includes 7 reference samples and the fourth sample set includes 10 reference samples, the ratio between the third sample number and the fourth sample number can be 7 / 10, i.e. 70%.

[0116] In the case where the predetermined threshold is 70%, it can be determined that the third sample set and the fourth sample set meet the predetermined condition, and thus the negative sample can be determined as the target negative sample.

[0117] According to an embodiment of the present disclosure, since the target negative sample is determined according to the ratio between the third sample number and the fourth sample number by the third sample number and the fourth sample number, the accuracy of determining the target negative sample is improved.

[0118] According to an embodiment of the present disclosure, the positive sample set includes a plurality of positive sample data, and the reference sample set includes a plurality of reference sample data; before determining the first sample set from the reference sample set, the sample determination method further includes: determining continuous target positive sample data of a continuous type from the plurality of positive sample data; determining continuous target first reference sample data matching the continuous target positive sample data from the plurality of reference sample data; determining discrete target positive sample data of a discrete type from the plurality of positive sample data; determining discrete target first reference sample data matching the discrete target positive sample data from the plurality of reference sample data; determining a first continuous sample similarity based on the continuous target positive sample data and the continuous target first reference sample data; determining a first discrete sample similarity based on the discrete target positive sample data and the discrete target first reference sample data; and determining a first target similarity between the positive sample and the reference sample based on the first continuous sample similarity and the first discrete sample similarity.

[0119] According to an embodiment of the present disclosure, the continuous target positive sample data of the continuous type can include asset data, transaction data, and stored amount data.

[0120] According to an embodiment of the present disclosure, the continuous target positive sample data of the continuous type can be determined from the plurality of positive sample data according to a predetermined continuous type. For example, in a case where the predetermined continuous type is asset data, asset data is determined as the continuous target positive sample data from the plurality of positive sample data.

[0121] According to an embodiment of the present disclosure, the asset data can be determined as the continuous target first reference sample data from the first reference sample data in a case where the continuous target positive sample data is determined as the asset data.

[0122] According to an embodiment of the present disclosure, the discrete target positive sample data of the discrete type can be determined from the plurality of positive sample data according to a predetermined discrete type.

[0123] According to an embodiment of the present disclosure, the discrete target positive sample data of the discrete type can include object type data. For example, in a case where the predetermined discrete type is object type data, object type data is determined as the discrete target positive sample data from the plurality of positive sample data.

[0124] According to an embodiment of the present disclosure, the object type data can be determined as the discrete target first reference sample data from the first reference sample data in a case where the discrete target positive sample data is determined as the object type data.

[0125] According to an embodiment of the present disclosure, the first continuous sample similarity can be determined by comparing the Euclidean distance between the feature vector of the continuous target positive sample data and the feature vector of the continuous target first reference sample data.

[0126] According to an embodiment of the present disclosure, the first discrete sample similarity can be determined by comparing the Euclidean distance between the feature vector of the discrete target positive sample data and the feature vector of the discrete target first reference sample data.

[0127] According to an embodiment of the present disclosure, the first target similarity can be obtained by adding the first continuous sample similarity and the first discrete sample similarity.

[0128] Thus, the first sample set can be determined from the reference sample set by the first target similarity.

[0129] According to an embodiment of the present disclosure, since the positive sample data and the first reference sample data are divided into discrete type and continuous type, the similarity is calculated for the positive sample data and the first reference sample data of different types, and then the first target similarity is determined according to the calculated first continuous sample similarity and the first discrete sample similarity, which improves the accuracy of the first target similarity.

[0130] Figure 5 An illustrative diagram of a method for determining a first target similarity according to an embodiment of the present disclosure is shown.

[0131] As shown in Figure 5 positive sample data 502 is determined from the positive sample 501, and the continuous target positive sample data 503 of the continuous type and the discrete target positive sample data 504 of the discrete type are determined from the positive sample data 502. Then the first reference sample data 506 is determined from the reference sample 505, and the discrete target first reference sample data 507 of the discrete type and the continuous target first reference sample data 508 of the continuous type are determined from the first reference sample data 506. Then the first discrete sample similarity 509 is determined according to the discrete target positive sample data 504 and the discrete target first reference sample data 507, and the first continuous sample similarity 510 is determined according to the continuous target positive sample data 503 and the continuous target first reference sample data 508 of the continuous type. The first target similarity 511 is determined according to the first discrete sample similarity 509 and the first continuous sample similarity 510.

[0132] According to an embodiment of the present disclosure, the first continuous sample similarity is determined based on the continuous target positive sample data and the continuous target first reference sample data, including: determining a first feature vector of the continuous target positive sample data; determining a second feature vector of the continuous target first reference sample data; and determining the first continuous sample similarity based on the first feature vector and the second feature vector.

[0133] According to an embodiment of the present disclosure, the first feature vector can be obtained by performing data analysis on the continuous target positive sample data, and the second feature vector can be obtained by performing data analysis on the continuous target first reference sample data.

[0134] According to an embodiment of the present disclosure, the first feature vector and the second feature vector are used to determine the first continuous sample similarity between the continuous target positive sample data and the continuous target first reference sample data. For example, the Euclidean distance between the first feature vector and the second feature vector is determined; and the first continuous sample similarity is determined according to the reciprocal of the Euclidean distance.

[0135] According to an embodiment of the present disclosure, since the first continuous sample similarity between the continuous target positive sample data and the continuous target first reference sample data is determined based on the first feature vector and the second feature vector, the first continuous sample similarity meeting the accuracy requirement is obtained, which helps to improve the accuracy of the first target similarity.

[0136] According to an embodiment of the present disclosure, the first discrete sample similarity is determined based on the discrete target positive sample data and the discrete target first reference sample data, including: in a case where the discrete target positive sample data and the discrete target first reference sample data are determined to be the same, determining the first discrete sample similarity based on the first continuous sample similarity; and in a case where the discrete target positive sample data and the discrete target first reference sample data are determined to be different, determining the first discrete sample similarity as a first predetermined similarity.

[0137] According to an embodiment of the present disclosure, the first discrete sample similarity is determined based on the first continuous sample similarity. In a case where there are a plurality of first continuous sample similarities between the positive sample and the reference sample, the average value of the plurality of first continuous sample similarities can be taken as the first discrete sample similarity. In a case where there is only one first continuous sample similarity between the positive sample and the reference sample, the value of the first discrete sample similarity can be obtained according to the value of the first continuous sample similarity.

[0138] According to an embodiment of the present disclosure, the first predetermined similarity can be 0, 0.1, 0.2, etc., which is not limited in the present disclosure.

[0139] According to the embodiment of the present disclosure, since the similarity between the discrete target positive sample data and the discrete target first reference sample data is determined in the case that the discrete target positive sample data and the discrete target first reference sample data are the same and different respectively, the first discrete sample similarity meeting the requirement of accuracy is obtained, which helps to improve the accuracy of the first target similarity.

[0140] According to the embodiment of the present disclosure, the negative sample set includes a plurality of negative sample data, and the reference sample set includes a plurality of reference sample data; before determining the first sample subset from the reference sample set, the sample determination method further includes: determining continuous target negative sample data of a continuous type from the plurality of negative sample data; determining continuous target second reference sample data matched with the continuous target negative sample data from the plurality of reference sample data; determining discrete target negative sample data of a discrete type from the plurality of negative sample data; determining discrete target second reference sample data matched with the discrete target negative sample data from the plurality of reference sample data; determining a second continuous sample similarity based on the continuous target negative sample data and the continuous target second reference sample data; determining a second discrete sample similarity based on the discrete target negative sample data and the discrete target second reference sample data; and determining a second target similarity between the negative sample and the reference sample based on the second continuous sample similarity and the second discrete sample similarity.

[0141] According to the embodiment of the present disclosure, the continuous target negative sample data of the continuous type can include asset data, transaction data, and stored amount data.

[0142] According to the embodiment of the present disclosure, the continuous target negative sample data of the continuous type can be determined from the plurality of negative sample data according to a predetermined continuous type. For example, in the case that the predetermined continuous type is asset data, the asset data is determined as the continuous target negative sample data from the plurality of negative sample data.

[0143] According to the embodiment of the present disclosure, the asset data can be determined as the continuous target second reference sample data from the second reference sample data in the case that the continuous target negative sample data is determined as the asset data.

[0144] According to the embodiment of the present disclosure, the discrete target negative sample data of the discrete type can be determined from the plurality of negative sample data according to a predetermined discrete type.

[0145] According to the embodiment of the present disclosure, the discrete target negative sample data of the discrete type can include object type data. For example, in the case that the predetermined discrete type is object type data, the object type data is determined as the discrete target negative sample data from the plurality of negative sample data.

[0146] According to an embodiment of the present disclosure, the second target similarity can be determined according to the second continuous sample similarity and the second discrete sample similarity.

[0147] According to an embodiment of the present disclosure, the second continuous sample similarity can be determined according to the Euclidean distance between the feature vector of the continuous target negative sample data and the feature vector of the continuous target second reference sample data.

[0148] According to an embodiment of the present disclosure, the second discrete sample similarity can be determined according to the Euclidean distance between the feature vector of the discrete target negative sample data and the feature vector of the discrete target second reference sample data.

[0149] According to an embodiment of the present disclosure, the second target similarity can be determined by adding the second continuous sample similarity and the second discrete sample similarity.

[0150] According to an embodiment of the present disclosure, the first sample subset can be determined from the reference sample set according to the second target similarity.

[0151] According to an embodiment of the present disclosure, since the negative sample data and the second reference sample data are divided into discrete types and continuous types, the similarity is calculated for different types of negative sample data and second reference sample data, and then the second target similarity is determined according to the calculated second continuous sample similarity and second discrete sample similarity, which improves the accuracy of the second target similarity.

[0152] Figure 6 An illustrative diagram of a method for determining a second target similarity according to an embodiment of the present disclosure is shown.

[0153] As Figure 6As shown, the negative sample data 602 is determined from the negative sample 601, and the continuous target negative sample data 603 of the continuous type and the discrete target negative sample data 604 of the discrete type are determined from the negative sample data 602. Then the second reference sample data 606 is determined from the reference sample 605, and the discrete target second reference sample data 607 of the discrete type and the continuous target second reference sample data 608 of the continuous type are determined from the second reference sample data 606. Then the second discrete sample similarity 609 is determined according to the discrete target negative sample data 604 and the discrete target second reference sample data 607, and the second continuous sample similarity 610 is determined according to the continuous target negative sample data 603 and the continuous target second reference sample data 608 of the continuous type. The second target similarity 611 is determined according to the second discrete sample similarity 609 and the second continuous sample similarity 610.

[0154] According to an embodiment of the present disclosure, the second continuous sample similarity is determined based on the continuous target negative sample data and the continuous target second reference sample data, including: determining a third feature vector of the continuous target negative sample data; determining a fourth feature vector of the continuous target second reference sample data; and determining the second continuous sample similarity based on the third feature vector and the fourth feature vector.

[0155] According to an embodiment of the present disclosure, the third feature vector can be obtained by data analysis on the continuous target negative sample data, and the fourth feature vector can be obtained by data analysis on the continuous target second reference sample data.

[0156] According to an embodiment of the present disclosure, the third feature vector and the fourth feature vector are used to determine the second continuous sample similarity between the continuous target negative sample data and the continuous target second reference sample data. For example, the Euclidean distance between the third feature vector and the fourth feature vector is determined; and the second continuous sample similarity is determined according to the reciprocal of the Euclidean distance.

[0157] According to an embodiment of the present disclosure, since the second continuous sample similarity between the continuous target negative sample data and the continuous target second reference sample data is determined according to the third feature vector and the fourth feature vector, the second continuous sample similarity meeting the accuracy requirement is obtained, which helps to improve the accuracy of the second target similarity.

[0158] According to an embodiment of the present disclosure, the second discrete sample similarity is determined based on the discrete target negative sample data and the discrete target second reference sample data, including: in a case where the discrete target negative sample data and the discrete target second reference sample data are determined to be the same, determining the second discrete sample similarity based on the second continuous sample similarity; and in a case where the discrete target negative sample data and the discrete target second reference sample data are determined to be different, determining the second discrete sample similarity as a second predetermined similarity.

[0159] According to an embodiment of the present disclosure, the second discrete sample similarity is determined based on the second continuous sample similarity. In a case where there are multiple second continuous sample similarities between the negative sample and the reference sample, the average of the multiple second continuous sample similarities can be taken as the second discrete sample similarity. In a case where there is only one second continuous sample similarity between the negative sample and the reference sample, the value of the second discrete sample similarity can be obtained according to the value of the second continuous sample similarity.

[0160] According to an embodiment of the present disclosure, the second predetermined similarity can be 0, 0.1, 0.2, etc., which is not limited in the present disclosure.

[0161] According to an embodiment of the present disclosure, since the similarity between the discrete target negative sample data and the discrete target second reference sample data is determined in the case where the discrete target negative sample data and the discrete target second reference sample data are the same and different, respectively, the second discrete sample similarity meeting the accuracy requirement is obtained, which helps to improve the accuracy of the second target similarity.

[0162] In order to better understand the content of the present disclosure, an embodiment of the present disclosure is given herein, for example: a plurality of unlabeled samples and a plurality of labeled samples can be obtained first, and the sample features of the obtained plurality of unlabeled samples and the plurality of labeled samples are processed to be the same, and the processing method is not limited in the present disclosure. Then, the plurality of unlabeled samples are taken as reference samples to construct a reference sample set M, and a labeled sample set K is constructed according to the plurality of labeled samples. The labeled sample set K can be divided into a training set T, a validation set V and a test set P. The positive sample set P and the negative sample set N can be obtained according to the labeled samples in the training set T. The number of reference samples in the reference sample set M can be greater than or equal to the number of labeled samples in the labeled sample set K. The unlabeled sample can be a sample to be determined a label, and the labeled sample can be a sample with a determined label. The label can be used to identify the attribute information of the labeled sample, which can be n-bit attribute code, and the form of the attribute code can be one-hot code, label code, etc., which is not limited in the present disclosure.

[0163] Then, the similarity between each positive sample in the positive sample set P and each reference sample in the reference sample set M is calculated to obtain a first similarity result set Similar_P, and the similarity between each negative sample in the negative sample set N and each reference sample in the reference sample set M is calculated to obtain a second similarity result set Similar_N. The similarities in the first similarity result set Similar_P and the second similarity result set Similar_N can be sorted respectively, and the first sample set and the third sample set can be obtained according to the top 10% to 50% similarities respectively.

[0164] If the negative sample data of the negative sample and the positive sample data of the positive sample are clear and explicit, and can provide classification information meeting the requirements, there should be no intersection or less intersection between the reference samples satisfying the similarity threshold determined for any negative sample in the negative sample set or any positive sample in the positive sample set, that is, the ratio between the first sample set and the second sample set should be lower than a predetermined threshold.

[0165] Based on this, the first sample set Similar_sample_Px of the positive sample in the reference sample set can be determined, and the first sample set Similar_sample_Px is used for similarity calculation with the reference samples in the first sample subsets to obtain a second sample set Similar_sample_Px_n.

[0166] The ratio between the first sample quantity num(Similar_sample_Px) of the first sample set and the second sample quantity num(Similar_sample_Px_n) of the second sample set is determined according to the first sample quantity num(Similar_sample_Px) and the second sample quantity num(Similar_sample_Px_n), and the higher the ratio is, the greater the possibility that the positive sample quality is low is.

[0167] The ratio between the first sample quantity and the second sample quantity can be calculated as follows: [num(Similar_sample_Px)] / [num(Similar_sample_Px_n)].

[0168] Therefore, the target positive sample can be determined according to the ratio.

[0169] For the negative sample set, the target negative sample can also be determined by the similar method, which will not be described here.

[0170] Figure 7 A flowchart of a model training method according to an embodiment of the present disclosure is schematically shown.

[0171] As Figure 7 shown, the model training method of this embodiment includes operation S710~operation S730.

[0172] In operation S710, according to the sample determination method of the present disclosure, a target positive sample is determined from the fifth sample set.

[0173] In operation S720, the target positive sample is removed from the fifth sample set to obtain a sixth sample set.

[0174] In operation S730, based on the sixth sample set, an initial model is trained to obtain a target model.

[0175] According to an embodiment of the present disclosure, the fifth sample set can be the training set T described above. Thus, the sixth sample set can be the training set L from which the target positive sample is removed.

[0176] The initial model can be trained using the training set L to obtain an intermediate model, and the parameter information of the intermediate model can be adjusted using the validation set V described above to obtain the target model.

[0177] The test set P described above can be input into the target model to determine whether the determined target positive sample is accurate according to the effect of the target model. If the target model is improved, it can be determined that the determined target positive sample is an accurate target positive sample. If it is not improved or decreased, the target positive sample can be re-determined.

[0178] In some embodiments, the target positive sample and the target negative sample can be removed from the training set T described above together, and the accuracy of the determined target positive sample and the target negative sample can be verified. Similar to the above method, details are not repeated here.

[0179] According to an embodiment of the present disclosure, based on the first sample set determined by the positive sample and the plurality of first sample subsets determined by the negative sample set, the second sample set is determined from the plurality of first sample subsets based on the first sample set, and then the positive sample is determined as the target positive sample according to the first sample set and the second sample set. The computing power is fully utilized, the purpose of automatically determining the target positive sample is achieved, the resources are saved, and the efficiency of determining the target positive sample is improved.

[0180] Based on the above sample determination method, the present disclosure further provides a sample determination device. The device will be described in detail below. Figure 8

[0181] Figure 8 The structure block diagram of the sample determination device according to an embodiment of the present disclosure is schematically shown.

[0182] As Figure 8 ​As shown, the sample determination apparatus 800 of this embodiment includes a first determination module 810, a second determination module 820, a third determination module 830, and a fourth determination module 840.

[0183] The first determination module 810 is configured to determine, for each positive sample in the positive sample set, a first sample set from the reference sample set, wherein the similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold. In an embodiment, the first determination module 810 can be configured to perform the operation S210 described above, and thus details are not repeated here.

[0184] The second determination module 820 is configured to determine, for each negative sample in the negative sample set, a first sample subset from the reference sample set, wherein the similarity between each reference sample in the first sample subset and the negative sample satisfies the similarity threshold. In an embodiment, the second determination module 820 can be configured to perform the operation S220 described above, and thus details are not repeated here.

[0185] The third determination module 830 is configured to determine, based on the first sample set, a second sample set from the plurality of first sample subsets. In an embodiment, the third determination module 830 can be configured to perform the operation S230 described above, and thus details are not repeated here.

[0186] The fourth determination module 840 is configured to determine, in a case where the first sample set and the second sample set satisfy a predetermined condition, that the positive sample is a target positive sample. In an embodiment, the fourth determination module 840 can be configured to perform the operation S240 described above, and thus details are not repeated here.

[0187] According to an embodiment of the present disclosure, the sample determination apparatus described above further includes a response determination module and a calling module. The response determination module is configured to determine, in response to a sample determination instruction from an electronic device, the positive sample set, the negative sample set, and the reference sample set from a sample database; and the calling module is configured to call a sample transmission interface to obtain the positive sample set, the negative sample set, and the reference sample set from the sample database.

[0188] According to an embodiment of the present disclosure, the sample determination apparatus described above is further configured to process the positive sample, the negative sample, and the reference sample. The positive sample includes positive sample data, and the positive sample data includes asset data, object type data, storage duration data, and transaction data. The negative sample includes negative sample data, and the negative sample data includes asset data, object type data, storage duration data, and transaction data. The reference sample includes reference sample data, and the reference sample data includes asset data, object type data, storage duration data, and transaction data.

[0189] According to an embodiment of the present disclosure, the sample determination apparatus further includes a sixth determination module, a seventh determination module, and an eighth determination module. The sixth determination module is configured to determine a first sample number of the first sample set. The seventh determination module is configured to determine a second sample number of the second sample set. The eighth determination module is configured to determine that the first sample set and the second sample set satisfy the predetermined condition in a case where a ratio between the first sample number and the second sample number is greater than a predetermined threshold.

[0190] According to an embodiment of the present disclosure, the sample determination apparatus further includes a ninth determination module, a tenth determination module, an eleventh determination module, and a twelfth determination module. The ninth determination module is configured to determine, for each negative sample in the negative sample set, a third sample set from the reference sample set, wherein a similarity between each reference sample in the third sample set and the negative sample satisfies a similarity threshold. The tenth determination module is configured to determine, for each positive sample in the positive sample set, a second sample subset from the reference sample set, wherein a similarity between each reference sample in the second sample subset and the positive sample satisfies the similarity threshold. The eleventh determination module is configured to determine, based on the third sample set, a fourth sample set from the plurality of second sample subsets. The twelfth determination module is configured to determine that the negative sample is the target negative sample in a case where the third sample set and the fourth sample set satisfy the predetermined condition.

[0191] According to an embodiment of the present disclosure, the sample determination apparatus further includes a thirteenth determination module, a fourteenth determination module, and a fifteenth determination module. The thirteenth determination module is configured to determine a third sample number of the third sample set. The fourteenth determination module is configured to determine a fourth sample number of the fourth sample set. The fifteenth determination module is configured to determine that the third sample set and the fourth sample set satisfy the predetermined condition in a case where a ratio between the third sample number and the fourth sample number is greater than a predetermined threshold.

[0192] According to an embodiment of the present disclosure, the sample determination apparatus further includes a sixteenth determination module, a seventeenth determination module, an eighteenth determination module, a nineteenth determination module, a twentieth determination module, a twenty-first determination module, and a twenty-second determination module. The sixteenth determination module is configured to determine continuous target positive sample data of a continuous type from the plurality of positive sample data. The seventeenth determination module is configured to determine continuous target first reference sample data matching the continuous target positive sample data from the plurality of reference sample data. The eighteenth determination module is configured to determine discrete target positive sample data of a discrete type from the plurality of positive sample data. The nineteenth determination module is configured to determine discrete target first reference sample data matching the discrete target positive sample data from the plurality of reference sample data. The twentieth determination module is configured to determine a first continuous sample similarity based on the continuous target positive sample data and the continuous target first reference sample data. The twenty-first determination module is configured to determine a first discrete sample similarity based on the discrete target positive sample data and the discrete target first reference sample data. The twenty-second determination module is configured to determine a first target similarity between the positive sample and the reference sample based on the first continuous sample similarity and the first discrete sample similarity.

[0193] According to an embodiment of the present disclosure, the twentieth determination module includes a first determination submodule, a second determination submodule, and a third determination submodule. The first determination submodule is configured to determine a first feature vector of the continuous target positive sample data. The second determination submodule is configured to determine a second feature vector of the continuous target first reference sample data. The third determination submodule is configured to determine the first continuous sample similarity based on the first feature vector and the second feature vector.

[0194] According to an embodiment of the present disclosure, the twenty-first determination module includes a fourth determination submodule and a fifth determination submodule. The fourth determination submodule is configured to determine the first discrete sample similarity based on the first continuous sample similarity in a case where the discrete target positive sample data and the discrete target first reference sample data are determined to be the same. The fifth determination submodule is configured to determine the first discrete sample similarity as a first predetermined similarity in a case where the discrete target positive sample data and the discrete target first reference sample data are determined to be different.

[0195] According to an embodiment of the present disclosure, the sample determination apparatus further comprises a twenty-third determination module, a twenty-fourth determination module, a twenty-fifth determination module, a twenty-sixth determination module, a twenty-seventh determination module, a twenty-eighth determination module and a twenty-ninth determination module. The twenty-third determination module is configured to determine continuous target negative sample data of a continuous type from the plurality of negative sample data. The twenty-fourth determination module is configured to determine continuous target second reference sample data matching the continuous target negative sample data from the plurality of reference sample data. The twenty-fifth determination module is configured to determine discrete target negative sample data of a discrete type from the plurality of negative sample data. The twenty-sixth determination module is configured to determine discrete target second reference sample data matching the discrete target negative sample data from the plurality of reference sample data. The twenty-seventh determination module is configured to determine a second continuous sample similarity based on the continuous target negative sample data and the continuous target second reference sample data. The twenty-eighth determination module is configured to determine a second discrete sample similarity based on the discrete target negative sample data and the discrete target second reference sample data. The twenty-ninth determination module is configured to determine a second target similarity between the negative sample and the reference sample based on the second continuous sample similarity and the second discrete sample similarity.

[0196] According to an embodiment of the present disclosure, the twenty-seventh determination module comprises a sixth determination submodule, a seventh determination submodule and an eighth determination submodule. The sixth determination submodule is configured to determine a third feature vector of the continuous target negative sample data. The seventh determination submodule is configured to determine a fourth feature vector of the continuous target second reference sample data. The eighth determination submodule is configured to determine the second continuous sample similarity based on the third feature vector and the fourth feature vector.

[0197] According to an embodiment of the present disclosure, the twenty-eighth determination module comprises a ninth determination submodule and a tenth determination submodule. The ninth determination submodule is configured to determine the second discrete sample similarity based on the second continuous sample similarity in a case where the discrete target negative sample data and the discrete target second reference sample data are determined to be the same. The tenth determination submodule is configured to determine the second discrete sample similarity as a second predetermined similarity in a case where the discrete target negative sample data and the discrete target second reference sample data are determined to be different.

[0198] According to an embodiment of the present disclosure, any of the first determining module 810, the second determining module 820, the third determining module 830 and the fourth determining module 840 can be combined in one module, or any of them can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules, and implemented in one module. According to an embodiment of the present disclosure, at least one of the first determining module 810, the second determining module 820, the third determining module 830 and the fourth determining module 840 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system on package, an application specific integrated circuit (ASIC), or any other reasonable way of hardware or firmware that can be integrated or packaged with a circuit, or any one of software, hardware and firmware or any appropriate combination of any of them. Alternatively, at least one of the first determining module 810, the second determining module 820, the third determining module 830 and the fourth determining module 840 can be at least partially implemented as a computer program module that can perform corresponding functions when executed.

[0199] Based on the above model training method, the present disclosure further provides a model training device. The following will be described in detail Figure 9 The device is described in detail.

[0200] Figure 9 The structure block diagram of the model training device according to an embodiment of the present disclosure is schematically shown.

[0201] As Figure 9 shown, the model training device 900 of the embodiment includes a fifth determining module 910, a removing module 920 and a training module 930.

[0202] The fifth determining module 910 is configured to determine a target positive sample from the fifth sample set according to the sample determination method of the present disclosure. In an embodiment, the fifth determining module 910 can be configured to perform the operation S710 described above, which will not be repeated here.

[0203] The removing module 920 is configured to remove the target positive sample from the fifth sample set to obtain a sixth sample set. In an embodiment, the removing module 920 can be configured to perform the operation S720 described above, which will not be repeated here.

[0204] The training module 930 is configured to train the initial model based on the sixth sample set to obtain a target model. In an embodiment, the training module 930 can be configured to perform the operation S730 described above, which will not be repeated here.

[0205] According to an embodiment of the present disclosure, any of the fifth determining module 910, the removing module 920 and the training module 930 can be combined in one module, or any of them can be split into multiple modules. Alternatively, at least part of the function of one or more of these modules can be combined with at least part of the function of other modules, and implemented in one module. According to an embodiment of the present disclosure, at least one of the fifth determining module 910, the removing module 920 and the training module 930 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system on package, an application specific integrated circuit (ASIC), or any other reasonable way of hardware or firmware that can be integrated or packaged with a circuit, or implemented in any one of software, hardware and firmware or in a proper combination of any of them. Alternatively, at least one of the fifth determining module 910, the removing module 920 and the training module 930 can be at least partially implemented as a computer program module that can perform corresponding functions when it is run.

[0206] Figure 10 A block diagram of an electronic device suitable for implementing the sample determining method or the model training method according to an embodiment of the present disclosure is schematically shown.

[0207] As shown in Figure 10 The electronic device 1000 according to an embodiment of the present disclosure includes a processor 1001 that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage portion 1008 into a random access memory (RAM) 1003. The processor 1001 can include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset, and / or a special-purpose microprocessor (such as an application specific integrated circuit (ASIC)), and the like. The processor 1001 can also include an on-board memory for cache use. The processor 1001 can include a single processing unit or a plurality of processing units for performing different actions of the method processes according to embodiments of the present disclosure.

[0208] In the RAM 1003, various programs and data required for the operation of the electronic device 1000 are stored. The processor 1001, the ROM 1002, and the RAM 1003 are connected to each other via the bus 1004. The processor 1001 performs various operations of the method flow according to the embodiments of the present disclosure by executing the programs in the ROM 1002 and / or the RAM 1003. It should be noted that the programs can also be stored in one or more memories other than the ROM 1002 and the RAM 1003. The processor 1001 can also perform various operations of the method flow according to the embodiments of the present disclosure by executing the programs stored in the one or more memories.

[0209] According to an embodiment of the present disclosure, the electronic device 1000 can further include an input / output (I / O) interface 1005, which is also connected to the bus 1004. The electronic device 1000 can further include one or more of the following components connected to the input / output (I / O) interface 1005: an input part 1006 including a keyboard, a mouse, etc.; an output part 1007 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage part 1008 including a hard disk, etc.; and a communication part 1009 including a network interface card such as a LAN card, a modem, etc. The communication part 1009 performs communication processing via a network such as the Internet. A drive 1100 is also connected to the input / output (I / O) interface 1005 as necessary. A removable medium 1101 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc. is mounted on the drive 1100 as necessary, so that a computer program read out therefrom is installed in the storage part 1008 as necessary.

[0210] The present disclosure also provides a computer readable storage medium, which can be included in the device / apparatus / system described in the above embodiments; or can exist separately without being assembled into the device / apparatus / system. The above computer readable storage medium carries one or more programs, when the one or more programs are executed, the method according to the embodiments of the present disclosure is implemented.

[0211] According to an embodiment of the present disclosure, the computer readable storage medium can be a nonvolatile computer readable storage medium, for example, can include, but is not limited to, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any appropriate combination thereof. In the present disclosure, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present disclosure, the computer readable storage medium can include one or more memories such as the ROM 1002 and / or the RAM 1003 described above and / or one or more memory chips other than the ROM 1002 and the RAM 1003.

[0212] Embodiments of the present disclosure also include a computer program product including a computer program containing program codes for executing the methods shown in the flowcharts. When the computer program product is run in a computer system, the program codes are used to cause the computer system to implement the item recommendation method provided by the embodiments of the present disclosure.

[0213] The above-described functions defined in the system / device of the embodiments of the present disclosure are performed when the computer program is executed by the processor 1001. According to an embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by computer program modules.

[0214] In one embodiment, the computer program can rely on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program can also be transmitted, distributed, and downloaded in the form of a signal on a network medium and installed and downloaded through the communication part 1009 and / or installed from the detachable medium 1101. The program codes contained in the computer program can be transmitted by any appropriate network medium, including but not limited to wireless, wired, etc., or any appropriate combination thereof.

[0215] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1009 and / or installed from the detachable medium 1101. When the computer program is executed by the processor 1001, the above-described functions defined in the system of the embodiments of the present disclosure are performed. According to an embodiment of the present disclosure, the system, device, apparatus, module, unit, etc. described above can be implemented by computer program modules.

[0216] According to embodiments of the present disclosure, program code of the computer programs provided by embodiments of the present disclosure can be written in any combination of one or more programming languages, and specifically, these computer programs can be implemented using high-level procedural and / or object-oriented programming language, and / or assembly / machine language. Programming languages include, but are not limited to, such as Java, C++, python, “C” language or similar programming languages. Program code can be executed entirely on a user computing device, partially on a user device, partially on a remote computing device, or entirely on a remote computing device or server. In the case involving a remote computing device, the remote computing device can be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, connected to the Internet through an Internet service provider).

[0217] The flow diagrams and the block diagrams in the drawings are illustrations of possible architectures, functions, and operations for systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0218] Those skilled in the art can understand that the features described in various embodiments of the present disclosure and / or claims can be combined or / and integrated, even if such combinations or integrations are not explicitly described in the present disclosure. In particular, the features described in various embodiments of the present disclosure and / or claims can be combined and / or integrated in various combinations, without departing from the spirit and teachings of the present disclosure. All such combinations and / or integrations are within the scope of the present disclosure.

[0219] The above describes embodiments of the present disclosure. However, these embodiments are merely for illustrative purposes, and are not intended to limit the scope of the present disclosure. Although each embodiment is described above separately, this does not mean that the measures in each embodiment cannot be used advantageously in combination. The scope of the present disclosure is defined by the appended claims and their equivalents. Those skilled in the art can make various substitutions and modifications without departing from the scope of the present disclosure, and these substitutions and modifications should all fall within the scope of the present disclosure.

Claims

1. A sample determination method, comprising: For each positive sample in the positive sample set, determine a first sample set from a reference sample set, wherein a similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold; For each negative sample in the negative sample set, determine a first sample subset from the reference sample set, wherein the similarity between each reference sample in the first sample subset and the negative sample satisfies the similarity threshold; Based on the first sample set, determining a second sample set from a plurality of the first sample subsets; and When it is determined that the first sample set and the second sample set meet a predetermined condition, the positive sample is determined to be a target positive sample.

2. The method according to claim 1, wherein Before determining the first sample set from the reference sample set, the method further includes: In response to a sample determination instruction from the electronic device, determining the positive sample set, the negative sample set, and the reference sample set from a sample database; A sample transmission interface is called to obtain the positive sample set, the negative sample set, and the reference sample set from the sample database.

3. The method according to claim 1, wherein The positive sample includes positive sample data, and the positive sample data includes: asset data, object type data, storage duration data, and transaction data; the negative sample includes negative sample data, and the negative sample data includes: the asset data, the object type data, the storage duration data, and the transaction data; the benchmark sample includes benchmark sample data, and the benchmark sample data includes: the asset data, the object type data, the storage duration data, and the transaction data.

4. The method according to claim 1, before determining that the positive sample is a target positive sample when the first sample set and the second sample set satisfy a predetermined condition, the method further comprises: determining a first number of samples in the first sample set; determining a second number of samples in the second sample set; In the case where it is determined that the ratio between the first sample quantity and the second sample quantity is greater than a predetermined threshold, it is determined that the first sample set and the second sample set meet the predetermined condition.

5. The method according to claim 1, further comprising: For each negative sample in the negative sample set, determine a third sample set from the reference sample set, wherein the similarity between each reference sample in the third sample set and the negative sample satisfies the similarity threshold; For each positive sample in the positive sample set, determining a second sample subset from the reference sample set, wherein a similarity between each reference sample in the second sample subset and the positive sample satisfies the similarity threshold; Based on the third sample set, determining a fourth sample set from a plurality of the second sample subsets; and When it is determined that the third sample set and the fourth sample set meet the predetermined condition, the negative sample is determined to be a target negative sample.

6. The method according to claim 5, further comprising: before determining that the negative sample is a target negative sample when the third sample set and the fourth sample set satisfy a predetermined condition: determining a third number of samples in the third sample set; determining a fourth number of samples in the fourth sample set; In a case where it is determined that the ratio between the third sample quantity and the fourth sample quantity is greater than a predetermined threshold, it is determined that the third sample set and the fourth sample set meet the predetermined condition.

7. The method according to claim 1, wherein The positive sample set includes a plurality of positive sample data, and the reference sample includes a plurality of reference sample data; Before determining the first sample set from the reference sample set, the method further includes: Determining continuous target positive sample data of continuous types from the plurality of positive sample data; Determining continuous target first reference sample data that matches the continuous target positive sample data from the plurality of reference sample data; Determining discrete target positive sample data of discrete types from the plurality of positive sample data; Determine, from the plurality of reference sample data, discrete target first reference sample data that matches the discrete target positive sample data; Determining a first continuous sample similarity based on the continuous target positive sample data and the continuous target first reference sample data; Determining a first discrete sample similarity based on the discrete target positive sample data and the discrete target first reference sample data; A first target similarity between the positive sample and the reference sample is determined based on the first continuous sample similarity and the first discrete sample similarity.

8. The method according to claim 7, wherein: The determining of the first continuous sample similarity based on the continuous target positive sample data and the continuous target first reference sample data includes: Determining a first eigenvector of the continuous target positive sample data; Determining a second eigenvector of the continuous target first benchmark sample data; The first continuous sample similarity is determined based on the first feature vector and the second feature vector.

9. The method according to claim 7, wherein: The determining of the first discrete sample similarity based on the discrete target positive sample data and the discrete target first reference sample data includes: When it is determined that the discrete target positive sample data and the discrete target first reference sample data are identical, determining the first discrete sample similarity based on the first continuous sample similarity; and When it is determined that the discrete target positive sample data and the discrete target first reference sample are different, the first discrete sample similarity is determined to be a first predetermined similarity.

10. The method according to claim 1, wherein The negative sample set includes a plurality of negative sample data, and the reference sample includes a plurality of reference sample data; Before determining the first sample subset from the reference sample set, the method further includes: Determining continuous target negative sample data of continuous types from the plurality of negative sample data; Determining continuous target second reference sample data matching the continuous target negative sample data from the plurality of reference sample data; Determining discrete target negative sample data of discrete types from the plurality of negative sample data; Determining discrete target second reference sample data that matches the discrete target negative sample data from the plurality of reference sample data; Determining a second continuous sample similarity based on the continuous target negative sample data and the continuous target second reference sample data; Determining a second discrete sample similarity based on the discrete target negative sample data and the discrete target second reference sample data; A second target similarity between the negative sample and the reference sample is determined based on the second continuous sample similarity and the second discrete sample similarity.

11. The method according to claim 10, wherein: The determining of the second continuous sample similarity based on the continuous target negative sample data and the continuous target second reference sample data includes: Determining a third eigenvector of the continuous target negative sample data; determining a fourth eigenvector of the second reference sample data of the continuous target; The second continuous sample similarity is determined based on the third eigenvector and the fourth eigenvector.

12. The method according to claim 10, wherein: The determining of the second discrete sample similarity based on the discrete target negative sample data and the discrete target second reference sample data includes: When it is determined that the discrete target negative sample data and the discrete target second reference sample data are the same, determining the second discrete sample similarity based on the second continuous sample similarity; and When it is determined that the discrete target negative sample data and the discrete target second reference sample are different, the second discrete sample similarity is determined to be a second predetermined similarity.

13. A model training method comprising: The sample determination method according to any one of claims 1 to 12, determining the target positive sample from a fifth sample set; removing the target positive sample from the fifth sample set to obtain a sixth sample set; Based on the sixth sample set, an initial model is trained to obtain a target model.

14. A sample determination device, comprising: A first determining module is configured to determine, for each positive sample in the positive sample set, a first sample set from a reference sample set, wherein a similarity between each reference sample in the first sample set and the positive sample satisfies a similarity threshold; A second determining module is configured to determine, for each negative sample in the negative sample set, a first sample subset from the reference sample set, wherein a similarity between each reference sample in the first sample subset and the negative sample satisfies the similarity threshold; A third determining module is configured to determine a second sample set from a plurality of first sample subsets based on the first sample set; and The fourth determining module is configured to determine that the positive sample is a target positive sample when it is determined that the first sample set and the second sample set meet a predetermined condition.

15. A model training device comprising: A fifth determination module, configured to determine the target positive sample from a fifth sample set according to the sample determination method according to any one of claims 1 to 12; a removal module, configured to remove the target positive sample from the fifth sample set to obtain a sixth sample set; The training module is used to train the initial model based on the sixth sample set to obtain the target model.

16. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are enabled to execute the method according to any one of claims 1 to 13.

17. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to perform the method according to any one of claims 1 to 13.

18. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 13 is implemented.

Images