Data processing method and apparatus, computing device, and test reduction device

Abstract

The embodiment of the present application provides a data processing method and device, a computing device and a test simplification device, the data processing method comprises the following steps: in response to a parameter optimization request, determining a target function corresponding to the parameter optimization request; in the process that the target function performs parameter test on any candidate parameter, calling a test simplification module to judge whether the parameter test satisfies a simplification condition, and obtaining a judgment result; if the judgment result is that the parameter test satisfies the simplification condition, stopping the parameter test of the candidate parameter; if the judgment result is that the parameter test does not satisfy the simplification condition, continuing to perform the parameter test of the candidate parameter, and obtaining a test result of the candidate parameter. The embodiment of the present application improves the parameter test efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and particularly relates to a data processing method and device, a computing device and a test simplification device. BACKGROUND

[0002] In the model calculation process of a machine learning model, a neural network model and the like, many parameters need to be involved, for example, hyperparameters constituted by parameters such as the number of iterations and the network depth, node parameters set for each computing node in the neural network model training process and the like. In order to obtain more accurate calculation results, the parameters of the model usually need to be set to an optimal set of parameters. The more common parameter optimization methods can mainly include a black-box optimization algorithm and a white-box optimization algorithm.

[0003] In the prior art, the objective function in the black-box optimization algorithm has a "black-box" feature, and the mathematical expression form of the objective function is unknown, which has high complexity. When the black-box optimization algorithm is used to optimize the parameters, a series of candidate parameters are generated, and the objective function is used to perform parameter tests on the multiple candidate parameters respectively to obtain test results corresponding to the multiple candidate parameters respectively, and then the target parameter with the optimal test result is selected from the multiple candidate parameters according to the test results.

[0004] As can be known from the above description, the existing parameter optimization method is to perform parameter tests on multiple candidate parameters respectively to obtain test results of all candidate parameters, and the parameter test process is relatively complex, which leads to low parameter optimization efficiency. SUMMARY

[0005] Therefore, the embodiments of the present application provide a data processing method and device, a computing device and a test simplification device to solve the technical problem of low parameter optimization efficiency caused by the complex parameter test process in the prior art.

[0006] In a first aspect, the embodiments of the present application provide a data processing method, comprising:

[0007] In response to a parameter optimization request, determining an objective function corresponding to the parameter optimization request;

[0008] In the process of performing a parameter test on any candidate parameter by the objective function, calling a test simplification module to determine whether the parameter test meets a simplification condition and obtaining a determination result;

[0009] If the determination result is that the parameter test meets the simplification condition, stopping the parameter test of the candidate parameter;

[0010] If the determination result is that the parameter test does not meet the simplification condition, continuing to perform the parameter test of the candidate parameter to obtain a test result of the candidate parameter.

[0011] In a second aspect, the embodiments of the present application provide a data processing method, comprising:

[0012] In response to a request of calling a parameter processing interface, determining a processing resource corresponding to the parameter processing interface;

[0013] Using the processing resource corresponding to the parameter processing interface to execute the following steps:

[0014] In response to a parameter optimization request, determining a target function corresponding to the parameter optimization request;

[0015] In a process of performing parameter test on any candidate parameter by the target function, calling a test simplification module to determine whether the parameter test meets a simplification condition, and obtaining a determination result;

[0016] If the determination result is that the parameter test meets the simplification condition, stopping the parameter test of the candidate parameter;

[0017] If the determination result is that the parameter test does not meet the simplification condition, continuing to perform the parameter test of the candidate parameter, and obtaining a test result of the candidate parameter.

[0018] In a third aspect, the embodiments of the present application provide a data processing method, comprising:

[0019] Receiving a determination request initiated by a computing device for determining whether a parameter test on any candidate parameter by a target function meets a simplification condition; wherein the target function is determined by the computing device in response to a parameter optimization request;

[0020] In response to the determination request, determining whether the parameter test meets the simplification condition;

[0021] The parameter test of the candidate parameter is stopped when the simplification condition is met, and the parameter test of the candidate parameter is continued to obtain a test result of the candidate parameter when the simplification condition is not met.

[0022] In a fourth aspect, the embodiments of the present application provide a data processing apparatus, comprising:

[0023] A first response module configured to determine a target function corresponding to a parameter optimization request in response to the parameter optimization request;

[0024] A result obtaining module configured to call a test simplification module to determine whether a parameter test on any candidate parameter by the target function meets a simplification condition, and obtain a determination result in a process of performing the parameter test on the candidate parameter by the target function;

[0025] The first processing module is configured to stop the parameter test of the candidate parameter if the result of the judgment indicates that the parameter test satisfies the simplified condition.

[0026] The second processing module is configured to continue the parameter test of the candidate parameter if the result of the judgment indicates that the parameter test does not satisfy the simplified condition, and obtain a test result of the candidate parameter.

[0027] In a fifth aspect, an embodiment of the present application provides a data processing apparatus, comprising:

[0028] The request receiving module is configured to receive a judgment request initiated by a computing device for whether a parameter test of a candidate parameter on a target function satisfies a simplified condition, wherein the target function is determined by the computing device in response to a parameter optimization request;

[0029] The second response module is configured to judge whether the parameter test satisfies the simplified condition in response to the judgment request.

[0030] The parameter test of the candidate parameter is stopped when the simplified condition is satisfied, and the parameter test of the candidate parameter is continued to obtain a test result of the candidate parameter when the simplified condition is not satisfied.

[0031] In a sixth aspect, an embodiment of the present application further provides a computing device, which can comprise a storage component and a processing component; the storage component is configured to store one or more computer instructions; the one or more computer instructions are invoked by the processing component;

[0032] The processing component can be configured to:

[0033] In response to a parameter optimization request, determine a target function corresponding to the parameter optimization request;

[0034] In a process of performing a parameter test on any candidate parameter on the target function, invoke a test simplification module to judge whether the parameter test satisfies a simplified condition, and obtain a judgment result;

[0035] If the judgment result indicates that the parameter test satisfies the simplified condition, stop the parameter test of the candidate parameter.

[0036] If the judgment result indicates that the parameter test does not satisfy the simplified condition, continue the parameter test of the candidate parameter, and obtain a test result of the candidate parameter.

[0037] In a seventh aspect, an embodiment of the present application further provides a test simplification device, which can comprise a storage component and a processing component; the storage component is configured to store one or more computer instructions; the one or more computer instructions are invoked by the processing component;

[0038] The processing component can be configured to:

[0039] receive a judgment request initiated by the computing device for whether a parameter test of a candidate parameter by a target function satisfies a simplified condition; wherein the target function is determined by the computing device in response to a parameter optimization request;

[0040] in response to the judgment request, judge whether the parameter test satisfies the simplified condition;

[0041] wherein the parameter test of the candidate parameter is stopped when the simplified condition is satisfied; and the parameter test of the candidate parameter is continued to obtain a test result of the candidate parameter when the simplified condition is not satisfied.

[0042] In the embodiments of the present application, during a parameter test of a candidate parameter by a target function, the parameter test of the candidate parameter is interrupted to obtain a test intermediate result of the candidate parameter. When the test intermediate result of the candidate parameter satisfies a simplified condition, the parameter test of the candidate parameter is stopped. When the test intermediate result of the candidate parameter does not satisfy the simplified condition, the parameter test of the candidate parameter is continued to obtain a test result of the candidate parameter. By monitoring the test intermediate result of the candidate parameter, it is determined whether the candidate parameter needs to continue the test. When the simplified condition is satisfied, the parameter test of the candidate parameter is stopped, which can improve the test efficiency of the candidate parameter, reduce unnecessary parameter test process of the candidate parameter, and improve the parameter optimization efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0044] Figure 1 A flowchart of one embodiment of a data processing method provided by the embodiments of the present application;

[0045] Figure 2 A flowchart of another embodiment of a data processing method provided by the embodiments of the present application;

[0046] Figure 3 A flowchart of another embodiment of a data processing method provided by the embodiments of the present application;

[0047] Figure 4 A flowchart of another embodiment of a data processing method provided by the embodiments of the present application;

[0048] Figure 5A flow chart of still another embodiment of the data processing method provided in the embodiments of the present application;

[0049] Figure 6 A flow chart of still another embodiment of the data processing method provided in the embodiments of the present application;

[0050] Figure 7 A flow chart of still another embodiment of the data processing method provided in the embodiments of the present application;

[0051] Figure 8 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application;

[0052] Figure 9 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application;

[0053] Figure 10 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application;

[0054] Figure 11 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application;

[0055] Figure 12 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application;

[0056] Figure 13 A schematic diagram of one embodiment of the data processing device provided in the embodiments of the present application. DETAILED DESCRIPTION

[0057] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0058] The terms used in the embodiments of the present application are only for the purpose of describing the specific embodiments, and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the embodiments of the present application and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Plural" generally includes at least two, but does not exclude the case of including at least one.

[0059] It should be understood that the term "and / or" as used herein merely describes an associated relationship between associated objects, and can represent three relationships, for example, A and / or B can represent three cases of A alone, A and B together, and B alone. In addition, the character " / " herein generally represents an "or" relationship between the front and rear associated objects.

[0060] Depending on the context, the word "if" as used herein can be interpreted as meaning "when" or "while" or "in response to determining" or "in response to identifying". Similarly, depending on the context, the phrase "if it is determined" or "if it is identified (a stated condition or event)" can be interpreted as meaning "when it is determined" or "in response to determining" or "when it is identified (a stated condition or event)" or "in response to identifying (a stated condition or event)".

[0061] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover non-exclusive inclusion, so that a product or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such product or system. Without more limitations, the element defined by the phrase "comprising a" does not exclude the presence of additional identical elements in the product or system comprising the element.

[0062] The technical solution of the embodiment of the present application can be applied to the parameter optimization scene. In the parameter optimization process, the test conditions of the candidate parameters are judged to determine whether to perform the subsequent parameter optimization process of the parameters, reduce the optimization process of the parameters, and improve the optimization efficiency of the parameters.

[0063] In the prior art, many parameters are involved in the calculation process of the machine learning model, neural network model and other models. The selection of the parameters often has an important influence on the calculation result of the model. Therefore, some parameter optimization algorithms can be designed, and the optimal parameters can be selected according to the set parameter optimization method. In the common black box optimization algorithm, the mathematical expression of the objective function is unknown, the specific calculation process of the objective function is unknown, but the calculation result of the parameters that can be optimized is known. In the actual process of optimizing the parameters by using the black box optimization algorithm, a series of candidate parameters can be generated, and the generated multiple candidate parameters can be subjected to parameter test by using the objective function to obtain multiple test results corresponding to the multiple candidate parameters respectively, and then the target parameter with the optimal test effect can be selected from the test results of the multiple candidate parameters respectively. However, this parameter optimization method needs to perform parameter test on each candidate parameter to obtain the test result of each candidate parameter. Since the process of parameter test is relatively complex, a large number of test calculations are actually required in the parameter optimization process, and the efficiency of parameter optimization is low.

[0064] In this embodiment, during the parameter testing of any candidate parameter in the objective function, the parameter testing of that candidate parameter is interrupted to obtain intermediate test results. If the intermediate test results of the candidate parameter meet the simplification condition, the parameter testing of that candidate parameter is stopped. If the intermediate test results of the candidate parameter do not meet the simplification condition, the parameter testing of that candidate parameter continues to obtain the test results. By monitoring the intermediate test results of candidate parameters to confirm whether the candidate parameter needs to continue participating in the test, and stopping the parameter testing of the candidate parameter if the simplification condition is met, the efficiency of candidate parameter testing can be improved, unnecessary candidate parameter testing processes can be reduced, and parameter optimization efficiency can be improved.

[0065] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0066] like Figure 1 The diagram shown is a flowchart of one embodiment of a data processing method provided in this application. The method may include the following steps:

[0067] 101: In response to the parameter optimization request, determine the objective function corresponding to the parameter optimization request.

[0068] The data processing method provided in this application can be applied to computing devices, such as computers, servers, cloud servers, supercomputers, laptops, tablets, etc. This application does not impose too many limitations on the specific type of computing device.

[0069] Optionally, the parameter experimentation process can involve calculating the model of the input candidate parameters for the objective function to obtain the output result, which is the target value of the candidate parameter. In the actual process of parameter optimization or screening, an iterative approach can be used to continuously sample candidate parameters and conduct parameter experiments on the sampled candidate parameters to obtain the target value corresponding to each candidate parameter. Finally, the target parameter with the optimal target value can be selected from the multiple candidate parameters after the parameter experiment has been completed.

[0070] The parameter optimization request may include experimental information for parameter experiments, such as function information of the objective function and parameter sampling strategies. Function information of the objective function may include, for example, function name, call link, or function identifier, which can identify different functions. The parameter sampling strategy may specifically be the method for generating candidate parameters, through which new candidate parameters can be continuously generated. The objective function may include mathematical models or network models, and may have "black box" characteristics, meaning that the mathematical expression of the objective function is unknown or is a highly complex computational model that is difficult to describe and characterize directly by mathematical formulas.

[0071] Optionally, the parameter optimization request can be initiated by a user, the user can provide the test information of the parameter test, the user terminal can detect the test information of the parameter test provided by the user, and generate the parameter optimization request based on the test information of the parameter test, and send the parameter optimization request to the computing device configured with the data processing method as shown in the Figure 1 In addition, the parameter optimization request can also be automatically generated by the computing device or other client when it is judged that parameter selection is needed, for example, when it is detected that the target function needs parameter selection, the parameter optimization request can be generated based on the target function and the parameter sampling strategy of the target function, and the parameter optimization request can be provided to the computing device configured with the data processing method as shown in the Figure 1

[0072] At this time, it can also include receiving the parameter optimization request sent by the user, and after obtaining the judgment result of the candidate parameter, the judgment result can be output to the user to realize the parameter optimization interaction with the user.

[0073] 102: In the process of the target function performing parameter test on any candidate parameter, the test simplification module is called to judge whether the parameter test meets the simplification condition, and the judgment result is obtained.

[0074] In the embodiment of the application, the candidate parameter can be a parameter that needs to be optimized in a mathematical calculation model such as a machine learning model, a neural network model, a three-dimensional calculation model, and a game model. The candidate parameter in the embodiment of the application can be a general model parameter, a hyperparameter, a game model parameter, a data model parameter, etc. The type and number of candidate parameters are not limited in the embodiment of the application. The hyperparameter can be a parameter set before learning, rather than a parameter in the model training process, and can not be involved in the actual training process. For example, the network depth of the machine learning model, the number of iterations, and the number of neurons in each layer can all belong to the hyperparameter. The blood loss value generated when the game program is attacked can also belong to the hyperparameter. In the e-commerce field, the number of search terms used when searching for terms can also belong to the hyperparameter. In the financial market, the time step of the market, the feature dimension, etc. also belong to the hyperparameter.

[0075] ​The candidate parameter can be a parameter value corresponding to a plurality of sub-parameters respectively, and the candidate parameter can also be referred to as a candidate parameter instance. For example, a certain candidate parameter can include parameter values corresponding to three sub-parameters A, B and C, such as when A takes 0.1, B takes 0.3, and C takes 0.1, a candidate parameter can be formed, and when A takes 0.1, B takes 0.3, and C takes 0.15, a candidate parameter can also be formed. The number and values of the sub-parameters of the candidate parameter can be set according to the actual use requirements of the parameter. As a possible implementation manner, the first candidate parameter can be randomly generated or obtained based on the input of the user, and the subsequent candidate parameters can be obtained by resampling based on the historical parameters and the test results of the historical parameters.

[0076] 103: If the judgment result is that the parameter test meets the simplification condition, stop the parameter test of the candidate parameter.

[0077] 104: If the judgment result is that the parameter test does not meet the simplification condition, continue to perform the parameter test of the candidate parameter, and obtain the test result of the candidate parameter.

[0078] The simplification condition refers to a condition judgment that can be performed on whether the candidate parameter needs to complete the entire parameter test. The simplification condition can be determined by the intermediate result of the test of the candidate parameter, so as to effectively monitor the test process of the candidate parameter, avoid invalid parameter test, and improve the parameter test efficiency.

[0079] In the embodiment of the application, in response to the parameter optimization request, after determining the target function corresponding to the parameter optimization request, the test simplification module can be called to judge whether the parameter test meets the simplification condition in the process of performing the parameter test on any candidate parameter by the target function, and obtain a judgment result. If the judgment result of the candidate parameter is that the candidate parameter meets the simplification condition, the parameter test of the candidate parameter is stopped. If the judgment result is that the parameter test does not meet the simplification condition, the parameter test of the candidate parameter can be continued to be performed, and the test result corresponding to the candidate parameter is obtained. By monitoring the parameter test of the candidate parameter, it is confirmed whether the candidate parameter needs to complete the entire parameter test, so as to improve the test efficiency of the candidate parameter, reduce unnecessary candidate parameter test process, and improve the parameter optimization efficiency.

[0080] Figure 1In the illustrated embodiments, the number of candidate parameters can include one or more. That is, parameter testing can be performed on one candidate parameter at a time. To improve testing efficiency, parameter testing can also be performed on multiple candidate parameters simultaneously. The judgment of simplified conditions can be applied to all one or more parameter tests to monitor the parameter testing process of one or more candidate parameters. Furthermore, when performing parameter testing on multiple candidate parameters simultaneously, a multi-process, hardware-software combined approach can be used to further improve parameter testing efficiency.

[0081] In practical applications, the technical solution of this application embodiment can be configured in a cloud server, and users can send parameter optimization requests to the configured cloud server through a user client. Figure 1 In the cloud server of the data processing method shown, after passing through the cloud server, the cloud server can execute... Figure 1 The data processing method shown in the diagram feeds back the obtained experimental results to the user terminal, which then outputs the experimental results of the candidate parameters to the user. After obtaining at least one candidate parameter that meets the simplification conditions, the experimental results corresponding to each of the at least one candidate parameter can be obtained. Based on the experimental results corresponding to each of the at least one candidate parameter, the target parameter that meets the parameter optimization conditions is selected from the at least one candidate parameter. The target parameter can then be sent to the user terminal, which outputs the optimal target parameter to the user.

[0082] As one possible implementation, during the parameter experiment of the objective function on any candidate parameter, intermediate experimental results of the candidate parameter can be obtained, and the simplified conditions of the parameter experiment can be judged by the intermediate experimental results.

[0083] Calling the test simplification module to check whether the parameter test meets the simplification conditions can specifically include: monitoring the parameter test process of the objective function on the candidate parameters, determining whether the objective function has run to the preset parameter test monitoring node, and if so, obtaining the intermediate test results of the candidate parameters at the monitoring node.

[0084] During the parameter experiment, the target simplification algorithm determines when the parameter experiment runs to the monitoring node and obtains the intermediate results. The parameter experiment can continue to run. When the result shows that the parameter experiment meets the simplification conditions, the parameter experiment of the candidate parameters can be stopped.

[0085] Monitoring nodes can be specifically set according to the experimental stage of the parameter experiment. In some embodiments, monitoring nodes can be set at each stage of the parameter experiment. The parameter experiment stage can include: early experimental stage, mid-term experimental stage, and late experimental stage. Each parameter experiment stage can be set according to the calculation process of the objective function. Specifically, the objective function can be divided into a first function, a second function, and a third function according to the calculation order. The calculation stage of the first function can be the early experimental stage, where the first monitoring node can be set at the calculation result of the first function or a monitoring node can be set in the middle of the calculation process of the first function; the calculation stage of the second function can be the mid-term experimental stage, where the calculation result of the third function is the experimental result, and no monitoring node needs to be set, or a monitoring node can be set in the middle of the calculation process of the third function. The specific monitoring nodes can be set according to the actual monitoring needs. By dividing the experiment into early, mid, and late experimental stages, targeted monitoring can be carried out for different experimental periods of the parameter experiment, achieving efficient and effective monitoring.

[0086] As another possible implementation, during the parameter testing of any candidate parameter by the objective function, obtaining the intermediate test results of the candidate parameter can include: when the current execution result meets the monitoring conditions during the parameter testing of any candidate parameter by the objective function, obtaining the intermediate test results of the candidate parameter. The monitoring conditions can be set according to the specific process of the experiment and the monitoring requirements. For example, the monitoring condition can be set to the iteration count reaching an iteration threshold.

[0087] In practical applications, the test simplification module can provide multiple test simplification algorithms, and can be the program or subroutine required to perform the function of judging whether the parameter test meets the simplification conditions. The test simplification module can also provide an interface to communicate with the external environment to realize the transmission of data or information.

[0088] The simplified test module can be configured to have Figure 1 The illustrated embodiment is either incorporated in a computing device or directly configured in a simplified device, which can be configured with... Figure 1 The computing devices used in the data processing methods provided in the illustrated embodiments are various devices. Specifically, the simplified device can be a computer, server, cloud server, supercomputer, laptop, tablet, etc. This application does not impose excessive limitations on the specific type of simplified device.

[0089] like Figure 2 The diagram shown is a flowchart of another embodiment of a data processing method provided in this application. The method may include the following steps:

[0090] 201: In response to the parameter optimization request, determine the objective function corresponding to the parameter optimization request.

[0091] The steps in the embodiments of the present application are the same as those in the embodiments shown in the above Figure 1 For the sake of brevity of description, the same steps in the embodiments of the present application will not be described here.

[0092] Optionally, when the parameter optimization request is initiated by the user, the target function can be automatically matched for the parameter optimization request initiated by the user, or the user can determine the target function. When the target function is determined by the user, an input interface of the target function can be provided, and the user terminal can detect the target function input by the user and transmit it to the computing device for executing the data processing method.

[0093] Optionally, before determining the objective function corresponding to the parameter optimization request in response to the parameter optimization request, the method can further include detecting the parameter optimization request triggered by the user for any black box optimization algorithm. Determining the objective function corresponding to the parameter optimization request in response to the parameter optimization request can include determining the objective function corresponding to the black box optimization algorithm selected by the user in response to the parameter optimization request.

[0094] 202: In the process of the target function performing parameter test on any candidate parameter, calling a target reduction algorithm in the plurality of test reduction algorithms of the test reduction module to determine whether the parameter test meets the reduction condition, and obtaining a determination result.

[0095] The test reduction strategy can include a plurality of test reduction algorithms, and a target reduction algorithm can be selected from the plurality of test reduction algorithms to determine whether the parameter test meets the reduction condition. Providing a plurality of test reduction algorithms can be matched with different parameter tests, so as to increase the selectability of the reduction algorithm and provide a wider selection space.

[0096] 203: If the determination result is that the parameter test meets the reduction condition, stop the parameter test of the candidate parameter.

[0097] 204: If the determination result is that the parameter test does not meet the reduction condition, continue to perform the parameter test of the candidate parameter, and obtain the test result of the candidate parameter.

[0098] In the embodiments of the present application, in response to the parameter optimization request, after the target function corresponding to the parameter optimization request is determined, the target simplification algorithm in the test simplification module can be called to determine whether the parameter test satisfies the simplification condition in the process of the parameter test of any candidate parameter, and a determination result is obtained. If the determination result of the candidate parameter is that the candidate parameter satisfies the simplification condition, the parameter test of the candidate parameter is stopped. By determining the target simplification algorithm, the parameter test can be accurately determined, and an accurate determination result is obtained. If the determination result is that the parameter test does not satisfy the simplification condition, the parameter test of the candidate parameter can be continued to obtain the test result corresponding to the candidate parameter. By monitoring the parameter test of the candidate parameter, it is confirmed whether the candidate parameter needs to perform the entire parameter test completely, so as to improve the test efficiency of the candidate parameter, reduce unnecessary candidate parameter test process, and improve the parameter optimization efficiency.

[0099] As an embodiment, the target simplification algorithm can be determined from the plurality of test simplification algorithms in the following manner: the target simplification algorithm matched with the parameter test is searched from the plurality of test simplification algorithms of the test simplification module.

[0100] The determination step of the target simplification algorithm can be performed on the side of the computing device, or on the side of the device configured to test the simplification module.

[0101] Optionally, in the process of the parameter test of any candidate parameter by the target function, the test simplification module is called to determine whether the parameter test satisfies the simplification condition, and a determination result is obtained, which can include: generating a determination request of whether the parameter test satisfies the simplification condition; sending the determination request to the test simplification module, so that the test simplification module responds to the determination request and determines whether the parameter test satisfies the simplification condition.

[0102] The specific generation manner of the determination request can include various manners, and different generation manners can correspond to different response manners, and several generation manners of the determination request will be introduced below.

[0103] The first generation manner can generate a test identifier or a test handle of the parameter test of the candidate parameter by the target function to generate the determination request, and the test simplification module can monitor the parameter test to obtain a test intermediate result of the parameter test, and determine the simplification condition based on the test intermediate result.

[0104] The second generation manner can obtain a test intermediate result of the parameter test in the process of the parameter test of any candidate parameter by the target function, and generate a determination request based on the test intermediate result. The test simplification module can determine the simplification condition based on the test intermediate result after obtaining the test intermediate result in the determination request.

[0105] In the third generation, the target simplification algorithm can generate a judgment request during the parameter test of the target function on any candidate parameter. After the test simplification module obtains the target simplification algorithm in the judgment request, the target simplification algorithm can be fed back. After the computing device obtains the target simplification algorithm in the judgment request, the target simplification algorithm can be used to judge the parameter test of the candidate parameter. Specifically, the test intermediate result of the candidate parameter can be judged.

[0106] When the target simplification algorithm is executed on the computing device side, as shown in FIG. 8, it is a flowchart of another embodiment of a data processing method provided by the embodiment of the present application. The method can include the following steps: Figure 3

[0107] 301: In response to the parameter optimization request, determine the target function corresponding to the parameter optimization request.

[0108] 302: Determine the target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms.

[0109] 303: In the process of the parameter test of the target function on any candidate parameter, call the target simplification algorithm in the test simplification module to judge whether the parameter test meets the simplification condition, and obtain the judgment result.

[0110] 304: If the judgment result is that the parameter test meets the simplification condition, stop the parameter test of the candidate parameter.

[0111] 305: If the judgment result is that the parameter test does not meet the simplification condition, continue to execute the parameter test of the candidate parameter, and obtain the test result of the candidate parameter.

[0112] Optionally, determining the target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms can include: displaying the plurality of test simplification algorithms to the user, and obtaining the target simplification algorithm selected by the user from the plurality of test simplification algorithms.

[0113] ​In the parameter test process of the target optimization function on any candidate parameter, the target simplification algorithm of the test simplification module can be called to determine whether the parameter test meets the simplification condition, and a determination result is obtained. Then, a plurality of test simplification algorithms are determined. By determining the target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms, the selectable test simplification algorithm is provided, which effectively guarantees the parameter simplification scheme, to ensure the adaptation to different parameter tests, effectively expand the applicable range of parameter simplification, and improve the use efficiency of parameter simplification. After ensuring the target simplification algorithm, whether the parameter test of the candidate parameter meets the simplification condition can be determined according to the target simplification algorithm. If yes, the parameter test of the candidate parameter is stopped. If no, the parameter test of the candidate parameter is continued, and a test result of the candidate parameter is obtained. By effectively determining the test condition of the candidate parameter, the invalid parameter test of the candidate parameter is reduced, the time loss of the parameter test is reduced, and the test efficiency of the parameter test is improved.

[0114] As an embodiment, finding the target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms of the test simplification module can include:

[0115] The first test information corresponding to the parameter test of the candidate parameter is determined.

[0116] The second test information respectively associated with the plurality of test simplification strategies is obtained.

[0117] The target test information matched with the first test information in the second test information respectively corresponding to the plurality of test simplification algorithms is queried.

[0118] The test simplification algorithm corresponding to the target test information is determined as the target simplification algorithm.

[0119] Optionally, the test simplification algorithm can be a strategy for simplifying the test process of the candidate parameter, that is, the test simplification algorithm can estimate the training effect of the candidate parameter in the parameter test process of the candidate parameter to determine whether the candidate parameter needs to continue the parameter test.

[0120] Optionally, the test information can be attribute information related to the parameter test, and the test information of different parameter tests can be defined by using test attribute information. The first test information can be attribute information of the parameter test of the candidate parameter. The second test information can be attribute information of the parameter test applicable to the corresponding test simplification algorithm.

[0121] The first test information can be node information of a test monitoring node defined for the parameter test of the candidate parameter, a parameter test type corresponding to the parameter test of the candidate parameter, and / or a parameter test stage corresponding to the parameter test of the candidate parameter.

[0122] The second test information can include one or more of node information of a test monitoring node to which the corresponding test simplification algorithm is applicable, a parameter test type, and a parameter test stage.

[0123] The node information of the test monitoring node can be a mapping relationship or a mapping target of the objective function corresponding to the function of the test monitoring node. A test simplification algorithm matched with the node information can be established, and a corresponding association relationship can be set for the node information and the test simplification algorithm set for the node information. The target simplification strategy matched with the node information of the test monitoring node can be queried through the association relationship. For example, the hyperparameter to be optimized is the network depth, the mapping relationship of the function corresponding to the monitoring node is the influence of the dimension and reliability of feature extraction, and it is assumed that the influence of the dimension of feature extraction is that as the network depth increases, the higher the dimension of the extracted feature, the higher the reliability of the extracted feature. The test simplification algorithm can be set to be that the dimension of the feature is less than a dimension threshold and the reliability of the feature is higher than a precision threshold.

[0124] A specific example is that it is assumed that the dimension threshold set by the test simplification algorithm is 500 and the reliability is 85%. If the candidate parameter is a network depth of 10 and the dimension of the extracted feature is 1000 and the reliability is 70%, the candidate parameter 10 does not meet the simplification condition at this time, and the parameter test of the candidate parameter 10 can be stopped. If the candidate parameter is 5, the dimension of the extracted feature is 500, and the reliability is 90%. At this time, the candidate parameter 5 can meet the simplification condition, and the parameter test of the candidate parameter 5 can be continued to finally obtain the test result.

[0125] For another example, the blood loss value in the game scene can be used as a parameter to be optimized. From the perspective of user game experience, too high blood loss value can lead to too low game time, and too high blood loss value can lead to reduced game interest. In order to obtain a blood loss value that can balance the game time and the game interest, the user can define the interruption information of the blood loss value when evaluating the use effect of the blood loss value, and set a target simplification algorithm for the interruption information.

[0126] For example, in an e-commerce scenario, when recommending content to a user, the target recommended content can be obtained using multiple sub-parameters to represent the user's browsing behavior, click habits, and other characteristics. The weighted sum of the feature information corresponding to each sub-parameter can be used as the search feature of the user's target recommended content, and the proportion of each sub-parameter can be used as a candidate parameter for optimization. During the parameter test of the candidate parameter, a preset ratio of multiple sub-parameters can be used as a candidate parameter. When performing parameter test on the candidate parameter, the detection feature of the user can be determined through the candidate parameter, and the target recommended content can be searched for the user based on the search feature. Then, the click rate of the user on the target recommended content is predicted to determine whether the ratio of the multiple sub-parameters corresponding to the current can be used as the final result. Generally, the above test process is complex and computationally intensive. In order to save computation, the target recommended content found based on the current search feature can be used as a monitoring node to perform a simplified condition judgment on the target recommended content. For example, the similarity between the target recommended content and the current search feature is judged. If the similarity is lower than a preset similarity threshold, it can be confirmed that the target recommended content obtained based on the current search feature may not be accurate, and the subsequent click rate prediction process does not need to be performed. At this time, the parameter test of the current candidate parameter is stopped, and the generation and parameter test of the next candidate parameter can continue, reducing unnecessary parameter test and improving parameter test efficiency.

[0127] Optionally, since the test simplification algorithm can include multiple algorithms, different algorithm identifiers can be used to distinguish between different test simplification algorithms. When the algorithm identifier of the test simplification algorithm is known, if the specific algorithm logic of the test simplification algorithm is also known, the test information suitable for each test simplification algorithm can be recorded in detail. Therefore, the method can further include generating second test information for multiple test simplification algorithms. Since the test information can include one or more of the node information of the test monitoring node, the parameter test type, and the parameter test stage, the second test information corresponding to the multiple test simplification algorithms can include the applicable node information, zero, one or more of the applicable parameter test type, and zero, one or more of the applicable parameter test stage.

[0128] Any candidate parameter can have a corresponding parameter test type. The corresponding simplification algorithm can be set according to the parameter test type. As one possible implementation, the first test information can include the parameter test type.

[0129] Optionally, querying the target test information in the second test information corresponding to the multiple test simplification algorithms that matches the first test information can include:

[0130] The target test information matching the parameter test type of the first test information is queried from the second test information corresponding to each of the plurality of test simplification algorithms.

[0131] The plurality of test simplification algorithms can respectively correspond to second test information. The second test information can be attribute information of a parameter test applicable to the test simplification algorithm, and specifically can be enumeration information of test attribute information applicable to the test simplification algorithm. The second test information of the parameter test type existing in the first test information can be found through the second test information corresponding to each of the plurality of test simplification algorithms, so as to obtain the target test information matching the first test information.

[0132] In a possible design, the parameter test type can include a serial test type and a parallel test type.

[0133] Optionally, the querying of the target test information matching the parameter test type of the first test information from the second test information corresponding to each of the plurality of test simplification algorithms can include:

[0134] If the parameter test type of the candidate parameter is the serial test type, the target test information of the serial test type existing in the second test information corresponding to each of the plurality of test simplification algorithms is determined; or

[0135] If the parameter test type of the candidate parameter is the parallel test type, the target test information of the parallel test type existing in the second test information corresponding to each of the plurality of test simplification algorithms is determined.

[0136] The determination of the target test information of the serial test type existing in the second test information corresponding to each of the plurality of test simplification algorithms can specifically include: querying the target test information of the serial test type existing in the second test information corresponding to each of the plurality of test simplification algorithms. Any test simplification algorithm corresponds to second test information, if the second test information of a certain test simplification algorithm contains the serial test type, it indicates that the test simplification algorithm is applicable to the parameter test of the serial test type, and the test simplification algorithm can be determined as the target simplification algorithm.

[0137] The determination of the target test information of the parallel test type existing in the second test information corresponding to each of the plurality of test simplification algorithms can specifically include: querying the target test information of the parallel test type existing in the second test information corresponding to each of the plurality of test simplification algorithms. Any test simplification algorithm corresponds to second test information, if the second test information of a certain test simplification algorithm contains the parallel test type, it indicates that the test simplification algorithm is applicable to the parameter test of the parallel test type, and the test simplification algorithm can be determined as the target simplification algorithm.

[0138] As a further possible implementation, the first test information can comprise a parameter test phase.

[0139] Optionally, the querying the target test information matched with the first test information from the second test information respectively corresponding to the plurality of test reduction algorithms can comprise:

[0140] The target test information matched with the parameter test phase of the first test information from the second test information respectively corresponding to the plurality of test reduction algorithms is queried.

[0141] In a possible design, the parameter test phase can comprise an early test phase, a middle test phase and a late test phase.

[0142] Optionally, the querying the target test information matched with the parameter test phase of the first test information from the second test information respectively corresponding to the plurality of test reduction algorithms can comprise:

[0143] If the parameter test phase of the candidate parameter is the early test phase, the target test information with the early test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is determined to exist; or,

[0144] If the parameter test phase of the candidate parameter is the middle test phase, the target test information with the middle test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is determined to exist; or,

[0145] If the parameter test phase of the candidate parameter is the late test phase, the target test information with the late test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is determined to exist.

[0146] The target test information with the early test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is determined to exist specifically can comprise: the target test information with the early test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is queried. Any test reduction algorithm corresponds to the second test information, if the second test information of a certain test reduction algorithm contains the early test phase, it indicates that this test reduction algorithm is applicable to the parameter test of the early test phase, and it can be determined that the test reduction algorithm is the target reduction algorithm.

[0147] The target test information with the middle test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is determined to exist specifically can comprise: the target test information with the middle test phase from the second test information respectively corresponding to the plurality of test reduction algorithms is queried. Any test reduction algorithm corresponds to the second test information, if the second test information of a certain test reduction algorithm contains the middle test phase, it indicates that this test reduction algorithm is applicable to the parameter test of the middle test phase, and it can be determined that the test reduction algorithm is the target reduction algorithm.

[0148] The target test information corresponding to the existence of the late test stage can be determined from the second test information corresponding to each of the plurality of test simplification algorithms, and can include: querying the target test information corresponding to the existence of the middle test stage from the second test information corresponding to each of the plurality of test simplification algorithms. Each of the test simplification algorithms corresponds to the second test information. If the second test information of a certain test simplification algorithm contains the late test stage, it indicates that this test simplification algorithm is suitable for the parameter test of the late test stage, and the test simplification algorithm can be determined as the target test simplification algorithm.

[0149] In addition, the serial test type and the parallel test type are listed as the parameter test types in the embodiments of the present application. In actual application, in addition to the serial test type and the parallel test type, other parameter test types can also be included, for example, a sampling test type in which sampling is performed on a plurality of candidate parameters and then parameter test is performed. The embodiments of the present application do not make too many limitations on the parameter test type.

[0150] The parameter test stage is also divided in the embodiments of the present application, and is specifically divided into early, middle and late test stages. In actual application, the division of the parameter test stage is based on the test time and the specific calculation process in the test process. The stage division manner in the embodiments of the present application is only exemplary, and should not constitute a specific limitation on the test stage division in the present application. Any stage division based on the test time and the test calculation content or process can belong to the stage division scheme protected by the embodiments of the present application.

[0151] As another possible test manner, the first test information can further include: a parameter test stage and a parameter test type.

[0152] Optionally, querying the target test information matching the first test information from the second test information corresponding to each of the plurality of test simplification algorithms can include:

[0153] The target test information matching the parameter test type and the parameter test stage at the same time is determined from the second test information corresponding to each of the plurality of test simplification algorithms.

[0154] In another possible design, the parameter test type can include a parallel test type and a serial test type, and the parameter test stage can include: an early test stage, a middle test stage and a late test stage.

[0155] The target test information matching the parameter test type and the parameter test stage at the same time is determined from the second test information corresponding to each of the plurality of test simplification algorithms.

[0156] The target simplification algorithm obtained through the parameter test stage and the parameter test type can meet the requirements of the parameter test type and the parameter test stage, and provide a target simplification algorithm that meets more simplification requirements.

[0157] As another example, the plurality of test simplification algorithms includes a custom simplification algorithm set by the target user.

[0158] The target simplification algorithm can also be determined from the plurality of test simplification algorithms in the following manner:

[0159] If the plurality of test simplification algorithms includes a custom simplification algorithm set by the target user, the custom simplification algorithm is determined as the target simplification algorithm.

[0160] The first test information can also include a custom identifier set by the target user for the parameter test of the candidate parameter. The custom identifier is used to identify whether the target user sets a custom simplification algorithm. If the custom identifier is true, it is directly confirmed that the plurality of test simplification algorithms includes a custom simplification algorithm set by the target user for the candidate parameter, and the custom simplification algorithm set by the target user can be used as the target simplification algorithm. If the custom identifier is false, it is determined that the target user does not set a custom algorithm, and at this time, the target test information matching the first test information can be queried from the second test information corresponding to the plurality of test simplification algorithms.

[0161] As an example, the method can further include:

[0162] Based on the custom simplification algorithm of the target user, the test simplification module is controlled to save the custom simplification algorithm.

[0163] Optionally, an input interface of a simplification algorithm can be provided, and the simplification algorithm input by the user in the interface can be a custom simplification algorithm. By providing a custom simplification algorithm, the applicability of the simplification algorithm can be enhanced, the simplification of a plurality of types of parameter tests can be applied, and the test simplification range is expanded. The input interface of the simplification algorithm can be displayed by a computing device configured with the data processing method provided in the embodiments of the present application for a user who needs to perform parameter optimization, or the simplification module can be provided to the user to obtain the custom simplification algorithm input by the user. After the computing device obtains the custom simplification algorithm, the custom simplification algorithm can be sent to the test simplification module for saving. The test simplification module detects the custom simplification algorithm input by the user, and the custom simplification algorithm can be saved.

[0164] In some embodiments, if the plurality of test simplification algorithms includes a custom simplification algorithm set by the target user, the custom simplification algorithm can be determined as the target simplification algorithm, which can further include:

[0165] If there is a custom reduction algorithm set by the target user in the plurality of test reduction algorithms, a prompt information that the custom reduction algorithm exists is generated;

[0166] The target user is prompted to confirm whether the custom algorithm is applicable to the parameter test;

[0167] If the target user performs a confirmation operation on whether the custom reduction algorithm is applicable to the parameter test, the custom reduction algorithm is determined as the target reduction algorithm.

[0168] The custom reduction algorithm is prompted to the user to ensure that the user can effectively monitor the reduction strategy of the parameter test, and the effective interaction of the reduction algorithm is realized.

[0169] In some embodiments, the candidate parameter can be a parameter that needs to be tested by a parameter test. The use effect of the candidate parameter can be evaluated by the parameter test, and the test result of the parameter can be obtained.

[0170] As an embodiment, in the process of the target function performing the parameter test on any candidate parameter, the test reduction module is called to judge whether the parameter test meets the reduction condition, and the judgment result can include: in the process of the target function performing the parameter test on any candidate parameter, the test reduction module is called to judge whether the intermediate test result of the parameter test meets the reduction condition, and the judgment result is obtained.

[0171] The test intermediate result of the parameter test is obtained, and the reduction condition of the test intermediate result is judged to confirm whether the candidate parameter can continue the subsequent parameter test.

[0172] In a possible design, a monitoring node of the parameter test of the candidate parameter can be set. The parameter test of the candidate parameter can set a plurality of monitoring nodes to monitor a plurality of nodes. In the process of the target function performing the parameter test on any candidate parameter, the test intermediate result of the candidate parameter can include: in the process of the target function performing the parameter test on any candidate parameter, the test intermediate result of the candidate parameter corresponding to at least one monitoring node can be obtained.

[0173] As a possible implementation manner, the test intermediate result of the candidate parameter meeting the reduction condition can be that: when any one of the at least one test intermediate result of the candidate parameter meets the reduction condition, the candidate parameter meets the reduction condition, and the parameter test of the candidate parameter can be stopped.

[0174] As a further possible implementation, the intermediate test result of the candidate parameter not satisfying the simplification condition can specifically be that none of the at least one intermediate test result of the candidate parameter satisfies the simplification condition, the candidate parameter does not satisfy the simplification condition, and the parameter test of the candidate parameter is continued to obtain a test result.

[0175] In actual applications, the candidate parameter can be any of a plurality of parameter test types. Common parameter test types can include a serial test type and a parallel test type. The serial test type can refer to generating one candidate parameter per test and performing parameter test on each generated candidate parameter. The parallel test type can refer to generating a plurality of candidate parameters per test and simultaneously performing parameter test on the plurality of candidate parameters. The candidate parameter in the embodiments of the present application can be a candidate parameter generated in the serial test type or a candidate parameter generated in the parallel test type.

[0176] In some embodiments, in the process of performing parameter test on any candidate parameter by the target function, a target simplification algorithm of the plurality of test simplification algorithms of the test simplification module is called to determine whether the intermediate test result of the parameter test satisfies the simplification condition, and a determination result is obtained.

[0177] In the process of performing parameter test on any candidate parameter by the target function, a target simplification algorithm of the plurality of test simplification algorithms of the test simplification module is called to determine whether the intermediate test result of the parameter test satisfies the simplification condition, and a determination result is obtained.

[0178] As shown in FIG. 4, another embodiment of a data processing method provided by the embodiments of the present application is a flowchart, which can include the following steps: Figure 4

[0179] 401: In response to a parameter optimization request, a target function corresponding to the parameter optimization request is determined.

[0180] 402: A target simplification algorithm matched with the parameter test is searched from the plurality of test simplification algorithms of the test simplification module.

[0181] 403: In the process of performing parameter test on any candidate parameter by the target function, a target simplification algorithm of the plurality of test simplification algorithms of the test simplification module is called to determine whether the intermediate test result of the parameter test satisfies the simplification condition, and a determination result is obtained.

[0182] 404: If the determination result is that the parameter test satisfies the simplification condition, the parameter test of the candidate parameter is stopped.

[0183] 405: If the determination result is that the parameter test does not satisfy the simplification condition, the parameter test of the candidate parameter is continued to obtain a test result of the candidate parameter. ​

[0184] In the embodiments of the present application, in response to the parameter optimization request, the target function corresponding to the parameter optimization request can be found from the plurality of test reduction algorithms of the plurality of test reduction modules that match the parameter test. By providing the selection of the plurality of test reduction algorithms, the use of the test reduction algorithm can be expanded, the application scenario of the test reduction algorithm can be improved, and multi-dimensional angle application can be realized. In the process of performing parameter test on any candidate parameter by the target function, the target reduction algorithm in the plurality of test reduction algorithms of the test reduction model can be called to determine whether the intermediate test result in the test reduction algorithm satisfies the reduction condition, and a determination result is obtained. If the determination result is that the parameter test satisfies the reduction condition, the parameter test of the candidate parameter can be stopped, and if the determination result is that the parameter test does not satisfy the reduction condition, the parameter test of the candidate parameter can be continued, and a test result of the candidate parameter is obtained. By accurately judging the intermediate test result of the candidate parameter according to the reduction condition, accurate judgment is realized.

[0185] The test reduction algorithm can be a method for estimating the effect of the intermediate result of the test of the candidate parameter. In actual application, a plurality of test reduction algorithms can be provided at the same time, and a target reduction algorithm that matches the current test information can be selected from the plurality of test reduction algorithms.

[0186] In some embodiments, the target reduction algorithm can include a historical estimation algorithm.

[0187] Optionally, the historical estimation algorithm determines whether the intermediate test result satisfies the reduction condition by the following method:

[0188] According to the historical intermediate results and the historical test results corresponding to the plurality of historical parameters, an estimated test result corresponding to the intermediate test result is estimated;

[0189] It is determined whether the estimated test result matches the result threshold;

[0190] If not, it is determined that the intermediate test result satisfies the reduction condition;

[0191] If yes, it is determined that the intermediate test result does not satisfy the reduction condition.

[0192] Optionally, according to the historical intermediate results and the historical test results corresponding to the plurality of historical parameters, the estimated test result corresponding to the intermediate test result can include:

[0193] The historical intermediate results of the plurality of historical parameters are obtained at the monitoring node interrupting the parameter test of the candidate parameter.

[0194] According to the test intermediate result and the historical intermediate results corresponding to the historical parameters respectively, and in combination with the historical test results corresponding to the historical parameters respectively, an estimated test result corresponding to the test intermediate result is estimated.

[0195] Optionally, the determination of whether the estimated test result matches the result threshold value can specifically include: determining whether the estimated test result is greater than the result threshold value; or determining whether the estimated test result is less than the result threshold value. The size relationship between the estimated test result and the result threshold value can be determined according to the specific type of the objective function of the parameter test.

[0196] In a possible design, the estimation of the estimated test result corresponding to the test intermediate result according to the test intermediate result and the historical intermediate results corresponding to the historical parameters respectively, and in combination with the historical test results corresponding to the historical parameters respectively, can specifically include: determining a mapping relationship between the intermediate result and the test result according to the historical intermediate results corresponding to the historical parameters respectively and the historical test results; and determining the estimated test result corresponding to the test intermediate result according to the mapping relationship between the intermediate result and the test result.

[0197] Further, optionally, the determination of the mapping relationship between the intermediate result and the test result according to the historical intermediate results corresponding to the historical parameters respectively and the historical test results can specifically include: performing curve fitting on the historical intermediate results corresponding to the historical parameters respectively to obtain an intermediate result curve; performing curve fitting on the historical test results corresponding to the historical parameters respectively to obtain a test result curve; and determining the mapping relationship between the intermediate result and the test result according to the test result curve and the intermediate result curve.

[0198] The estimated test result corresponding to the test intermediate result is obtained through the result estimation method, so as to quickly determine whether the test intermediate result meets the simplification condition by using the estimated test result, and improve the accuracy of result estimation.

[0199] In some embodiments, the target simplification algorithm can further include: calculating a comparison algorithm.

[0200] The comparison algorithm specifically determines whether the test intermediate result meets the simplification condition in the following manner:

[0201] A reference intermediate value corresponding to a monitoring node that obtains the test intermediate result in the parameter test process is determined.

[0202] It is determined whether the reference intermediate value meets a preset reference threshold value.

[0203] If not, it is determined that the test intermediate result meets the simplification condition.

[0204] If yes, it is determined that the test intermediate result does not meet the simplification condition.

[0205] Optionally, intermediate reference values ​​can be determined based on historical test results of multiple historical parameters at monitoring nodes. Specifically, this can be achieved through weighted summation, mean calculation, variance calculation, etc.

[0206] The intermediate results can be compared with the mean of intermediate reference values ​​to determine whether they meet the simplification criteria. Specifically, determining whether the intermediate reference value meets a preset reference threshold can include: whether the difference between the intermediate reference value and the intermediate result is less than a preset difference threshold; if so, the intermediate result meets the simplification criteria; otherwise, it does not. Alternatively, in some embodiments, it can also be determined that the difference is greater than a preset difference threshold; if so, the intermediate result meets the simplification criteria; otherwise, it does not. The specific determination can be made based on actual usage requirements.

[0207] Alternatively, the variance of the intermediate experimental results can be calculated based on the intermediate reference value and the mean of the intermediate results. The variance value measures the stability of the intermediate results or the magnitude of the difference between the actual result and the mean. A larger variance value indicates greater stability, while a smaller variance value indicates less stability and greater deviation from the mean. Specifically, determining whether the intermediate reference value meets the preset reference threshold can include: whether the intermediate variance is less than the variance threshold; if so, the intermediate results meet the simplification conditions; if not, they do not. Alternatively, it can be determined whether the intermediate variance is greater than the variance threshold; if so, the intermediate results meet the simplification conditions; if not, they do not. The specific determination can be made based on actual usage requirements.

[0208] Optionally, both the historical estimation algorithm and the computational comparison algorithm can be part of a multi-trial simplification algorithm, and both can correspond to second trial information. When the historical estimation algorithm or the computational comparison algorithm is the target simplification algorithm, it can be determined that the second trial information of the historical estimation algorithm matches the first trial information corresponding to the parameter trial of the candidate parameter, or it can be determined that the second trial information of the computational comparison algorithm matches the first trial information corresponding to the parameter trial of the candidate parameter. The second trial information corresponding to the historical estimation algorithm and the second trial information of the computational comparison algorithm can be confirmed according to the specific trial attribute information of the corresponding parameter trial.

[0209] like Figure 5 The diagram shown is a flowchart of another embodiment of a data processing method provided in this application. The method may include the following steps:

[0210] 501: In response to the parameter optimization request, determine the target function corresponding to the parameter optimization request.

[0211] Some steps in the embodiments of the present application are the same as some steps in the foregoing embodiments, and for the sake of brevity of description, they will not be described here.

[0212] 502: In the process of parameter test of the target function on any candidate parameter, call the test simplification module to determine whether the parameter test meets the simplification condition, and obtain the determination result.

[0213] 503: If the determination result is that the parameter test meets the simplification condition, stop the parameter test of the candidate parameter.

[0214] 504: If the determination result is that the parameter test does not meet the simplification condition, continue to perform the parameter test of the candidate parameter, and obtain the test result of the candidate parameter.

[0215] 505: Determine at least one candidate parameter that does not meet the simplification condition, and obtain the test result corresponding to each of the at least one candidate parameter.

[0216] 506: According to the test result corresponding to each of the at least one candidate parameter, select a target parameter that meets the parameter optimization condition from the at least one candidate parameter.

[0217] In the embodiments of the present application, in the process of parameter test of the target function on any candidate parameter, the test intermediate result of the candidate parameter is obtained. If the test intermediate result of the candidate parameter meets the simplification condition, the parameter test of the candidate parameter can be stopped, and if the test intermediate result of the candidate parameter does not meet the simplification condition, the parameter test of the candidate parameter is continued to obtain the test result of the candidate parameter. By judging and verifying the parameter test of the candidate parameter, it is determined whether to continue to perform the parameter test of the candidate parameter, which can reduce unnecessary parameter test of the candidate parameter and reduce the time loss of the parameter test of the candidate parameter. Thus, at least one candidate parameter that does not meet the simplification condition can be determined, and the test result corresponding to each of the at least one candidate parameter is obtained. According to the test result corresponding to each of the at least one candidate parameter, a target parameter that meets the parameter optimization condition is selected from the at least one candidate parameter. At least one candidate parameter that meets the simplification condition is used as the basis for parameter selection, which reduces the complexity of the parameter space and improves the selection efficiency and effectiveness of the parameter.

[0218] Optionally, the test result information can be generated for any candidate parameter. When the objective function performs the parameter test on any candidate parameter, a test identifier can be generated for the candidate parameter, by which the parameter test of different candidate parameters can be identified. In addition, since the parameter test process of the candidate parameter is interrupted in the embodiments of the present application, after the judgment of the test condition, the candidate parameter can stop participating in the parameter test or can continue to participate in the parameter test. In order to distinguish the candidate parameters that complete the parameter test from the candidate parameters that do not complete the parameter test, the test state information can be set for the candidate parameter. The test state information can include: test in progress, completed or simplified, wherein test in progress indicates that the parameter test of the candidate parameter is still in progress, completed can identify that the candidate parameter has completed the parameter test and has the corresponding test result, and simplified can identify that the candidate parameter has not completed the parameter test and actually does not have the test result. Therefore, in order to distinguish the parameter test, the test result and the simplified result of different candidate parameters, the method can further include: generating first test information for the parameter test of the candidate parameter. The test result information of the candidate parameter can include: test identifier, candidate parameter, test result and test state information.

[0219] When the test state information of any candidate parameter is simplified, a poor test result can be directly set for the candidate parameter to directly distinguish the test result obtained by the candidate parameter that completes the parameter test. For example, assuming that the test result obtained by the candidate parameter that completes the parameter test is between 0.5 and 0.95, a test result of 0.01 to 0.1 can be set for the candidate parameter whose test state information is simplified, so as to more intuitively distinguish the candidate parameters with different test state information.

[0220] Optionally, determining at least one candidate parameter that does not satisfy the simplification condition and obtaining the test result corresponding to each of the at least one candidate parameter can include: obtaining the test result information corresponding to each of the plurality of candidate parameters, determining at least one candidate parameter whose test state information is completed from the test result information corresponding to each of the plurality of candidate parameters, and obtaining the test result in the test result information corresponding to each of the at least one candidate parameter.

[0221] The test result can be a use effect value obtained by the parameter evaluation of the parameter test on the candidate parameter. The higher the use effect value, the better the test result, and the lower the use effect value 2, the worse the test result. Selecting a target parameter that satisfies the parameter optimal condition from the at least one candidate parameter can include: selecting the candidate parameter with the largest use effect value corresponding to the test result from the test result corresponding to each of the at least one candidate parameter as the target parameter that satisfies the parameter optimal condition.

[0222] The parameter optimization problem can be directly related to many application fields. In order to improve the parameter optimization efficiency, the technical solution of the embodiment of the present application can be used.

[0223] In the distribution process of power resources and water resources, the distribution results of the power resources or the water resources in each region can be taken as a to-be-processed parameter to initiate a parameter optimization request. The to-be-processed parameter can be a resource amount corresponding to each region, for example, the load capacity of the region in the power scenario.

[0224] As an embodiment, before determining the target function corresponding to the parameter optimization request in response to the parameter optimization request, the following can also be included:

[0225] receiving a parameter optimization request initiated by a to-be-processed parameter for a target resource;

[0226] The determining of the target function corresponding to the parameter optimization request in response to the parameter optimization request includes:

[0227] In response to the parameter optimization request, a target function corresponding to a processing target of the target resource is determined.

[0228] Further including:

[0229] sampling the to-be-processed parameter multiple times to obtain multiple candidate parameters;

[0230] After the target parameter satisfying the parameter optimization condition is selected from the at least one candidate parameter according to the test results corresponding to the at least one candidate parameter, the following can also be included:

[0231] According to the value of the to-be-processed parameter in the target parameter, processing information of the target resource is generated to process the target resource according to the processing information.

[0232] The resource element represented by the to-be-processed parameter can be determined according to the processing target of the target resource. For example, when the processing target of the target resource can be the power load capacity set for different regions to make the total energy consumption of the power grid optimal, at this time, the power load capacity of different regions can be the to-be-processed parameter, and the calculation function of the total energy consumption of the power grid can be the processing target. The target parameter can be the power load capacity of each region in the case of obtaining the optimal total energy consumption of the power grid. According to the value of the to-be-processed parameter in the target parameter, the processing information of the target resource can be generated, that is, the prompt information or setting instruction of the power load capacity of each region can be generated according to the value of the to-be-processed parameter in the target parameter. The setting instruction can be used to set the capacity according to the power load capacity of each region. The prompt information can be displayed to the user to set the capacity of each region according to the power load capacity of each region prompted in the prompt information.

[0233] In the field of electronic commerce, parameter optimization problems can also be involved. For example, in the case of product recommendation, different users have different consumption habits, areas of interest, and historical browsing behaviors, and thus different browsing characteristics. In actual applications, in order to improve the click rate of users, the browsing characteristics of users, such as consumption habits and areas of interest, can be parameterized to generate different browsing parameters, and the characteristics of the click target of users can be accurately analyzed by setting multiple browsing parameters, so as to find a target product with higher user interest. The scheme of sampling multiple browsing parameters and performing parameter tests to determine the click probability of users can be applied to the technical scheme of the embodiments of the present application to improve the test efficiency.

[0234] Therefore, as an embodiment, before determining the objective function corresponding to the parameter optimization request in response to the parameter optimization request, the following can also be included:

[0235] Detecting a browsing operation initiated by a target user, and generating a parameter optimization request for a browsing parameter of the target user;

[0236] The determining of the objective function corresponding to the parameter optimization request in response to the parameter optimization request includes determining the objective function corresponding to the access target of the target user in response to the parameter optimization request.

[0237] Further comprising:

[0238] Sampling the browsing parameter multiple times to obtain multiple candidate parameters;

[0239] After selecting the target parameter that meets the parameter optimization condition from the at least one candidate parameter according to the test results corresponding to the at least one candidate parameter, the following can also be included:

[0240] Generating access recommendation information of the target user according to the value of the target parameter of the browsing parameter;

[0241] Finding a target product matching the access recommendation information from a product database to output the target product for the target user.

[0242] By setting the browsing parameter according to the browsing characteristics of the target user, the corresponding browsing parameter is obtained, and the parameter test of the browsing parameter is performed to obtain the optimal target parameter. Of course, in actual applications, the browsing parameter includes one or more sub-parameters, and multiple sub-parameters can form a candidate parameter when the sampling of the parameter values corresponding to the multiple sub-parameters is completed. After the target parameter is determined, the access recommendation information of the target user is determined, and thus the target product matching the access recommendation information can be found and output for the target user.

[0243] In some embodiments, the browsing parameter can be the proportion of different browsing features, and according to the value of the browsing parameter in the target parameter, the proportion of different browsing features in the product searching process can be determined, so that the multiple browsing features are weighted and summed according to their values in the target parameter to obtain the recommended features, which can be the access recommendation information.

[0244] In yet some embodiments, the browsing parameter can be the proportion of different types of products, that is, the products of multiple types can be recommended to the user respectively, but the proportions of the products of different types are different. Taking the makeup and clothing products as the main recommended types of products as an example, the recommendation proportions of the makeup products and the clothing products are parameter optimized, and finally the target parameter is 3:7, the makeup products are 3, and the clothing products are 7. At this time, according to the value of the browsing parameter in the target parameter, the generated access recommendation information can be searching for the makeup products and the clothing products according to the proportion of 3:7, that is, 3 makeup products and 7 clothing products matched with the recommendation information are searched from the product database and output to the user.

[0245] It should be noted that the specific application scheme of the parameter optimization shown in the present application is only exemplary and should not constitute a limitation on the application of the embodiments of the present application. The embodiments of the present application can be applied to various data model, calculation model parameter optimization scenarios.

[0246] In a possible design, the technical scheme of the embodiments of the present application can be configured in a server to form a service that can provide parameter optimization externally. As shown in Figure 6 The flowchart of another embodiment of a data processing method provided by the embodiments of the present application can include the following steps:

[0247] 601: In response to a request of calling a parameter processing interface, determining a processing resource corresponding to the parameter processing interface.

[0248] Using the processing resource corresponding to the parameter processing interface to execute the following steps:

[0249] 602: In response to a parameter optimization request, determining a target function corresponding to the parameter optimization request.

[0250] Before determining the target function corresponding to the parameter optimization request in response to the parameter optimization request, it further includes obtaining the parameter optimization request.

[0251] 603: In the process of parameter testing of the target function on any candidate parameter, calling a test simplification module to determine whether the parameter testing meets a simplification condition and obtaining a determination result.

[0252] 604: If the result of the judgment is that the parameter test satisfies the simplification condition, stop the parameter test of the candidate parameter.

[0253] 605: If the result of the judgment is that the parameter test does not satisfy the simplification condition, continue to perform the parameter test of the candidate parameter, and obtain the test result of the candidate parameter.

[0254] In a possible design, the processing resource corresponding to the parameter processing interface can further perform the following processing steps:

[0255] determining at least one candidate parameter that does not satisfy the simplification condition, and obtaining a test result corresponding to each of the at least one candidate parameter; and selecting a target parameter that satisfies a parameter optimization condition from the at least one candidate parameter according to the test result corresponding to each of the at least one candidate parameter.

[0256] In a possible design, the processing resource corresponding to the parameter processing interface can further perform the following processing steps:

[0257] obtaining a plurality of test simplification algorithms in response to a request of invoking the test simplification interface;

[0258] determining a target simplification algorithm that matches the parameter test from the plurality of test simplification algorithms.

[0259] Whether the test intermediate result of the candidate parameter satisfies the simplification condition is determined in the following manner:

[0260] determining, according to the target simplification algorithm, whether the test intermediate result satisfies the simplification condition.

[0261] In yet another possible design, the test simplification module includes a plurality of test simplification algorithms. The target simplification algorithm is determined from the plurality of test simplification algorithms in the following manner:

[0262] finding, from the plurality of test simplification algorithms of the test simplification module, the target simplification algorithm that matches the parameter test.

[0263] The specific steps performed by the processing resource corresponding to the parameter processing interface in the embodiments of the present application are the same as the processing steps performed by the data processing method shown in Figures 1-5 The specific implementation manners and technical effects of various technical features have been described in detail in the embodiments shown in Figures 1-5 Therefore, no further description is provided here.

[0264] As Figure 7 shown, a flowchart of another embodiment of a data processing method provided by the embodiments of the present application is shown. The method can include the following steps:

[0265] 701: Receive a judgment request initiated by a computing device for whether a parameter test of any candidate parameter on a target function satisfies a simplification condition.

[0266] wherein the objective function is determined by the computing device in response to a parameter optimization request.

[0267] 702: in response to the judgment request, judging whether the parameter test meets the simplification condition.

[0268] wherein the parameter test of the candidate parameter is stopped when the simplification condition is met; and the parameter test of the candidate parameter is continued to obtain the test result of the candidate parameter when the simplification condition is not met.

[0269] In the embodiments of the present application, when a judgment request initiated by the computing device for judging whether the parameter test of the candidate parameter meets the simplification condition of the objective function is received, the judgment request can be responded to, and whether the parameter test meets the simplification condition can be judged. By judging the parameter test according to the simplification condition, the accuracy of the parameter test can be improved.

[0270] As an embodiment, whether the parameter test meets the simplification condition is judged by the following method:

[0271] determining a target simplification algorithm matched with the parameter test from a plurality of test simplification algorithms;

[0272] judging whether the parameter test meets the simplification condition according to the target simplification algorithm, and generating a judgment result.

[0273] In some embodiments, the target simplification algorithm is determined from the plurality of test simplification algorithms by the following method: searching for the target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms of the test simplification module.

[0274] As another possible implementation, judging whether the parameter test meets the simplification condition according to the target simplification algorithm, and generating a judgment result can include: judging whether the intermediate test result of the parameter test meets the simplification condition based on the target simplification algorithm, and generating a judgment result.

[0275] Optionally, it further includes: receiving a judgment request of whether the parameter test of the candidate parameter meets the simplification condition of the objective function; wherein the objective function is determined by the computing device in response to a parameter optimization request. In response to the judgment request, whether the parameter test meets the simplification condition is judged, and a judgment result is generated. The judgment result includes that the parameter test meets the simplification condition or the parameter test does not meet the simplification condition; the parameter test of the candidate parameter is stopped when the simplification condition is met, and the parameter test of the candidate parameter is continued to obtain the test result of the candidate parameter when the simplification condition is not met.

[0276] Optionally, the step of determining whether the parameter test satisfies the simplification condition in response to the determination request can include: in response to the determination request, obtaining a target simplification algorithm provided by the computing device, and determining whether the parameter test satisfies the simplification condition based on the target simplification algorithm.

[0277] Optionally, the step of determining whether the parameter test satisfies the simplification condition in response to the determination request can include: in response to the determination request, searching for a target simplification algorithm matched with the parameter test from a plurality of test simplification algorithms of the test simplification module, and determining whether the parameter test satisfies the simplification condition based on the target simplification algorithm.

[0278] Optionally, the method further includes: in response to the determination request, obtaining an intermediate test result of the candidate parameter; and determining whether the intermediate test result satisfies the simplification condition. The step of determining whether the intermediate test result satisfies the simplification condition includes: determining a target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms; determining whether the intermediate test result satisfies the simplification condition according to the target simplification algorithm, and generating a determination result.

[0279] Figure 7 Some steps in the embodiments shown are the same as some steps in the embodiments shown in Figure 1 For the sake of brevity of description, the specific implementation modes and technical effects of the steps will not be described here.

[0280] For the sake of understanding, the parameter to be optimized is taken as an example of the network depth, the number of iterations, and the number of neurons in each layer of the machine learning model, and the cloud server provides a data processing method. An application example of an embodiment of the present application is described in detail by taking an example of interaction with a user device.

[0281] Reference is made to Figure 8 In actual application, a user device, for example, a terminal device such as a mobile phone terminal or an Internet of Things (IoT) terminal, can interact with a user, and the user device can communicate with a server capable of optimizing the hyperparameters. Taking a mobile phone terminal M1 as the user device and a cloud server M2 as the server as an example. The mobile phone terminal M1 can detect 801 a parameter optimization request of the hyperparameters of the network depth, the number of iterations, and the number of neurons in each layer of the machine learning model triggered by the user. The mobile phone terminal M1 can send 802 the parameter optimization request to the cloud server M2. The cloud server M2 can determine 803 a target function in response to the parameter optimization request.

[0282] Afterwards, the cloud server M2 can call 804 the test simplification module to determine whether the parameter test meets the simplification condition in the process of the target function performing parameter test on any candidate parameter, and obtain a determination result. If the determination result is that the parameter test meets the simplification condition, the parameter test of the candidate parameter is stopped 805. If the determination result is that the parameter test does not meet the simplification condition, the parameter test of the candidate parameter is continued 806 to obtain a test result of the candidate parameter. By interrupting the parameter test of the candidate parameter, the time consumption of the parameter test can be reduced.

[0283] In some embodiments, the cloud server M2 can also determine at least one candidate parameter that does not meet the simplification condition, and obtain 807 a test result corresponding to each of the at least one candidate parameter. According to the test result corresponding to each of the at least one candidate parameter, a target parameter that meets the parameter optimization condition is selected 808 from the at least one candidate parameter.

[0284] If the cloud server M2 performs selection of the target parameter that meets the parameter optimization condition, the target parameter finally selected can be sent 809 to the user terminal M1. After the user terminal M1 receives the target parameter, the target parameter can be displayed 810. The output mode of the target parameter can include various forms, for example, data, pages, information or messages, etc. The specific output mode of the target parameter is not limited in the embodiments of the present application.

[0285] In actual applications, the target parameter obtained by the parameter optimization method provided in the embodiments of the present application can also be directly applied to the model training scene of the machine learning model. For example, when the parameter optimized by the user is a hyperparameter, the optimal target parameter obtained is a target hyperparameter. The machine learning model can be constructed by using the target hyperparameter, and the machine learning model can be trained by using the training data to obtain the model parameters of the machine learning model constructed by using the target hyperparameter, so that the use effect of the machine learning model is better, for example, in the field of face recognition, the recognition accuracy of the face recognition model constructed by using the optimal hyperparameter is higher.

[0286] The technical solutions of the embodiments of the present application can be applied to artificial intelligence interaction, data retrieval, content recommendation, click rate prediction, intelligent factory, industrial control and other fields, especially in the field of content recommendation, such as content recommendation in the fields of e-commerce, video live broadcast, social field, online education, and resource allocation field, such as financial product configuration, power resource, water resource, supply chain allocation and other fields.

[0287] In order to facilitate understanding, the embodiments of the present application are described in detail with the following problem cases in several actual fields as examples.

[0288] (1) E-commerce field. In the e-commerce field, feature search, product recommendation in live broadcast scenarios, content recommendation, and computing advertising click rates are the most common application scenarios. In this embodiment, a content recommendation scenario is taken as an example for instance deployment. The general recommendation process in the recommendation scenario can be that the elements of the selected scenario are parameterized to obtain a plurality of parameters that affect the scenario, the plurality of parameters are used to identify different features of the scenario, and a candidate parameter is obtained by assigning features to the plurality of parameters. Based on the specific application requirements of the scenario, a target function matched with the scenario is selected, and when a user initiates a parameter optimization request for the candidate parameter, the candidate parameter is input into the target function for parameter testing. In the process of parameter testing of any candidate parameter by the target function, the test simplification module is called to determine whether the parameter testing meets the simplification condition, and a determination result is obtained. If the determination result is that the parameter testing meets the simplification condition, the parameter testing of the candidate parameter is stopped; if the determination result is that the parameter testing does not meet the simplification condition, the parameter testing of the candidate parameter is continued, and a test result of the candidate parameter is obtained.

[0289] Taking a click word recommendation scenario as an example, when a user clicks a search box in an APP (Application, application), the system will recommend some search words (Query words) to the user. The purpose of recommending search words to the user is to tap the user's potential purchase demand, increase the user's use stickiness, and improve the total number of commodity transactions. The search system uses the following architecture, combined with a deep learning Encode-Decode (encoder-decoder) network, i.e., a target network, to predict the recommendation of search words. It is assumed that the number of search words is optimized for parameter selection. In the prior art, the optimal parameter value of the parameter constituted by the number of search words is manually set according to artificial experience. Using the data processing method of the present patent, parameter testing of the number of search words can be automatically performed according to the above parameter optimization process. Then at least one search word quantity that meets the simplification condition is obtained, and at least one search word quantity respectively corresponding to the test result is obtained; according to the test result corresponding to at least one search word quantity respectively, at least one search word quantity that meets the parameter optimization condition is selected from at least one search word quantity. The obtained target search word quantity can be used as the optimal search word quantity. By interrupting the monitoring of the parameter testing of the number of search words, invalid testing of the current number of search words can be avoided, and the optimization efficiency of the number of search words can be improved.

[0290] (2) Social field, in the social field, the content recommendation is carried out to the social user, the material recommendation is carried out to the student and is also more common. The recommendation of social field is usually, social user browses social application program, and the display interface of application program shows the social content of user's interest. Usually, the recommendation of social field is usually, the historical browsing behavior of user, the field of attention, user information and the like option constitute characteristic parameter, and the combination of different options can constitute different parameters. When parameter is determined, namely, feature information can be generated, and based on the feature information of user's attention, the content related to the feature information is found. The content related to the feature information can be taken as a target function, in order to find the content of interest of social user, the number and kind of parameters can be optimized to obtain accurate social user content.

[0291] The technical scheme of the embodiment of the application can be configured in a cloud server, and the parameter optimization request can be initiated by an operation and maintenance personnel. The operation and maintenance personnel can set a plurality of parameters and a plurality of parameters constituted by the related information of the user, and then continuously generate candidate parameters, and then perform parameter test on each candidate parameter. In the process of parameter test on any candidate parameter by the target function, the test simplification module is called to judge whether the parameter test meets the simplification condition, and a judgment result is obtained. For example, the number of searches of the parameter test can be monitored. If the number of contents similar to the current set feature information reaches more than 60% when searching 1000 contents exceeds the preset number threshold, it can be determined that the parameter test does not meet the simplification condition, and if it does not exceed, it can be determined that the parameter test meets the simplification condition.

[0292] If the judgment result is that the parameter test meets the simplification condition, the parameter test of the candidate parameter is stopped; if the judgment result is that the parameter test does not meet the simplification condition, the parameter test of the candidate parameter is continued, and a test result of the candidate parameter is obtained. The parameter test is simplified by using the test simplification module, which can improve the parameter test efficiency and further ensure that the optimal feature option of the user is quickly obtained.

[0293] (3) Financial field. Stock index simulation is a very important problem. Based on linear regression, SVM (support vector machines) and LSTM (Long Short-Term Memory) and other models, the simulation of stock index is more common. Before using the model, a suitable model needs to be established. During the model training process, many hyperparameters can be involved, such as time_step (time step) in LSTM, feature_dim (feature dimension), hiddenfeatrue (hidden feature), and the like. Moreover, macro factors, micro factors, and unexpected events of the market are also involved. These context features will affect the selection of parameters. Using the patent technology, the candidate parameters of each parameter value can be obtained after setting the parameter values of each parameter multiple times, and the parameter test of each candidate parameter is performed to select the optimal target parameter from the candidate parameters.

[0294] In order to improve the parameter optimization efficiency of the above candidate parameters, the target function can be used to perform parameter test on the candidate parameters formed after the above hyperparameters are valued. In the process of performing parameter test on the candidate parameters using the target function, the test simplification module is called to determine whether the parameter test meets the simplification condition, and a determination result is obtained. If the determination result is that the parameter test meets the simplification condition, the parameter test of the candidate parameter is stopped. If the determination result is that the parameter test does not meet the simplification condition, the parameter test of the candidate parameter is continued, and a test result of the candidate parameter is obtained. At least one candidate parameter meeting the simplification condition is obtained, and at least one test result corresponding to the at least one candidate parameter is obtained. According to the test result corresponding to the at least one candidate parameter, a target parameter meeting the parameter optimization condition is selected from the at least one candidate parameter. The obtained target parameter can be used to construct a machine learning model corresponding to the index simulation problem, perform model training, and obtain model parameters. Then, the machine learning model obtained by training is used to simulate the RMSE (Root Mean Squared Error) difference value and other data of the actual stock index of the index simulation problem.

[0295] (4) Resource allocation field. Taking power resource allocation as an example. The allocation of power resources usually involves many regions, and each region can be represented by corresponding parameters. These parameters can allocate a certain proportion of resources, and the allocation of resources will affect regional economic, population, environmental information, and the like.

[0296] The technical scheme of the embodiment of the present application can be applied to the problems of dynamic pricing of the power market and economic load distribution of the power market. In the following, the specific application field of the power system is described in detail.

[0297] In the dynamic pricing problem of the electricity market, user type and electricity consumption are crucial factors affecting the market. These parameters can be used as optimization parameters. Sampling these parameters yields candidate parameters. The candidate parameters can then be used as the final optimization objective for the revenue / cost ratio in the power system, determining the corresponding objective function. Using this objective function, parameter experiments can be conducted on the candidate parameters to obtain the target revenue / cost value. By continuously experimenting with the candidate parameters, the optimal target value can be obtained. During the parameter experimentation process, the technical solution of this application can be applied to simplify the experimentation of candidate parameters, reducing the number of experiments with candidate parameters that yield poor predictions and improving the optimization efficiency.

[0298] In the problem of power load allocation, power suppliers can simultaneously provide power resources to multiple regions. The power load capacity of each region can be used as a candidate parameter, and the total energy consumption of the power grid can be used as the output of the objective function. Using the technical solution of this application embodiment, the power load capacity of each of the multiple regions can be set to obtain a candidate parameter. Then, parameter experiments are conducted on this candidate parameter using the objective function to obtain the calculation result of the candidate parameter. The objective function can represent a nonlinear constraint relationship between the power load capacity and the total energy consumption of the power grid. It can be represented as a black-box optimization algorithm to obtain the black-box objective function. Subsequently, by continuously conducting parameter experiments on the power load capacity of each region, the optimal load capacity allocation strategy is obtained. When conducting parameter experiments on the power load capacity of each region, the technical solution of this application embodiment can be used to simplify the parameter experiments, reducing the number of parameter experiments on candidate parameters with poor prediction effects and improving experimental efficiency.

[0299] like Figure 9 The diagram shown is a structural schematic of one embodiment of a data processing apparatus provided in this application. The apparatus may include:

[0300] First Response Module 901: Used to respond to parameter optimization requests and determine the objective function corresponding to the parameter optimization request.

[0301] Result Acquisition Module 902: Used to call the Experiment Simplification Module to determine whether the parameter experiment meets the simplification conditions during the parameter experiment of any candidate parameter in the objective function, and obtain the judgment result.

[0302] First processing module 903: If the judgment result is that the parameter experiment meets the simplification condition, then stop the parameter experiment of the candidate parameter.

[0303] The second processing module 904 is configured to, if the judgment result is that the parameter test does not satisfy the simplification condition, continue to perform the parameter test of the candidate parameter, and obtain a test result of the candidate parameter.

[0304] In the embodiment of the present application, in response to the parameter optimization request, after the target function corresponding to the parameter optimization request is determined, the test simplification module can be called to determine whether the parameter test satisfies the simplification condition in the process of performing the parameter test on any candidate parameter, and a judgment result is obtained. If the judgment result of the candidate parameter is that the candidate parameter satisfies the simplification condition, the parameter test of the candidate parameter is stopped. If the judgment result is that the parameter test does not satisfy the simplification condition, the parameter test of the candidate parameter can be continued, and a test result corresponding to the candidate parameter is obtained. The parameter test of the candidate parameter is monitored to determine whether the candidate parameter needs to be completely executed in the entire parameter test, so as to improve the test efficiency of the candidate parameter, reduce unnecessary candidate parameter test process, and improve the parameter optimization efficiency.

[0305] As an embodiment, the test simplification module includes a plurality of test simplification algorithms.

[0306] The result obtaining module can include:

[0307] The apparatus can further include a result obtaining unit configured to, in the process of performing the parameter test on any candidate parameter by the target function, call a target simplification algorithm in the plurality of test simplification algorithms of the test simplification module to determine whether the parameter test satisfies the simplification condition, and obtain a judgment result.

[0308] In some embodiments, the apparatus further includes:

[0309] An algorithm matching module is configured to search for a target simplification algorithm matched with the parameter test from the plurality of test simplification algorithms of the test simplification module.

[0310] As a possible implementation manner, the algorithm matching module can include:

[0311] A first determining unit is configured to determine first test information corresponding to the parameter test of the candidate parameter.

[0312] A first obtaining unit is configured to obtain second test information respectively associated with the plurality of test simplification algorithms.

[0313] An information matching unit is configured to query target test information matched with the first test information from the second test information respectively corresponding to the plurality of test simplification algorithms.

[0314] An algorithm determining unit is configured to determine that the test simplification algorithm corresponding to the target test information is the target simplification algorithm.

[0315] As a possible implementation manner, the first test information comprises: a parameter test type.

[0316] The information matching unit can comprise:

[0317] The first query sub-unit is configured to query, from the second test information respectively corresponding to the plurality of test reduction algorithms, target test information matching the parameter test type of the first test information.

[0318] In a possible design, the parameter test type comprises: a serial test type and a parallel test type.

[0319] The first query sub-unit can be specifically configured to:

[0320] If the parameter test type of the candidate parameter is the serial test type, the target test information of the serial test type is determined to exist in the second test information respectively corresponding to the plurality of test reduction algorithms; or,

[0321] If the parameter test type of the candidate parameter is the parallel test type, the target test information of the parallel test type is determined to exist in the second test information respectively corresponding to the plurality of test reduction algorithms.

[0322] As another possible implementation manner, the first test information comprises: a parameter test order.

[0323] The information matching unit can comprise:

[0324] The second query sub-unit is configured to query, from the second test information respectively corresponding to the plurality of test reduction algorithms, target test information matching the parameter test stage of the first test information.

[0325] In a possible design, the parameter test stage comprises: an early test stage, a middle test stage and a late test stage.

[0326] The second query sub-unit can be specifically configured to:

[0327] If the parameter test stage of the candidate parameter is the early test stage, the target test information of the early test stage is determined to exist in the second test information respectively corresponding to the plurality of test reduction algorithms; or,

[0328] If the parameter test stage of the candidate parameter is the middle test stage, the target test information of the middle test stage is determined to exist in the second test information respectively corresponding to the plurality of test reduction algorithms; or,

[0329] If the parameter test stage of the candidate parameter is the late test stage, the target test information of the late test stage is determined to exist in the second test information respectively corresponding to the plurality of test reduction algorithms.

[0330] As a further possible implementation, the first test information can further include: a parameter test stage and a parameter test type.

[0331] The information matching unit can include:

[0332] The third query sub-unit is configured to determine, from the second test information respectively corresponding to the plurality of test reduction algorithms, target test information that matches both the parameter test type and the parameter test stage.

[0333] In some embodiments, the result obtaining module can include:

[0334] The second obtaining unit is configured to, in the process of the parameter test on the candidate parameter by the target function, call the test reduction module to determine whether the intermediate test result of the parameter test satisfies the reduction condition, and obtain the determination result.

[0335] The second obtaining unit can be further configured to, in the process of the parameter test on the candidate parameter by the target function, call a target reduction algorithm in the plurality of test reduction algorithms of the test reduction module to determine whether the intermediate test result of the parameter test satisfies the reduction condition, and obtain the determination result.

[0336] The target reduction algorithm includes a historical estimation algorithm.

[0337] The second obtaining unit can include:

[0338] The result estimation sub-unit is configured to estimate, according to the historical intermediate result and the historical test result respectively corresponding to the plurality of historical parameters, an estimated test result corresponding to the test intermediate result.

[0339] The first determination sub-unit is configured to determine whether the estimated test result matches the result threshold value; if not, it is determined that the test intermediate result satisfies the reduction condition; if so, it is determined that the test intermediate result does not satisfy the reduction condition.

[0340] In some embodiments, the target reduction algorithm includes a calculation comparison algorithm.

[0341] The second obtaining unit can include:

[0342] The reference obtaining sub-unit is configured to determine an intermediate reference value corresponding to a monitoring node that obtains the test intermediate result in the parameter test process.

[0343] The second determination sub-unit is configured to determine whether the intermediate reference value satisfies a preset reference threshold value; if not, it is determined that the test intermediate result satisfies the reduction condition; if so, it is determined that the test intermediate result does not satisfy the reduction condition.

[0344] In a possible design, the plurality of test reduction algorithms includes a custom reduction algorithm set by a target user.

[0345] The algorithm matching module can comprise:

[0346] An algorithm matching unit is configured to determine the custom simplification algorithm as the target simplification algorithm if the custom simplification algorithm set by the target user exists in the plurality of trial simplification algorithms.

[0347] The apparatus can further comprise an algorithm saving module configured to control the trial simplification module to save the custom simplification algorithm based on the custom simplification algorithm of the target user.

[0348] Optionally, the algorithm matching unit can comprise:

[0349] A prompt generation sub-unit is configured to generate prompt information indicating that the custom simplification algorithm exists if the custom simplification algorithm set by the target user exists in the plurality of trial simplification algorithms.

[0350] An algorithm display sub-unit is configured to display the prompt information to the target user for the target user to confirm whether the custom algorithm is applicable to the parameter trial.

[0351] An algorithm determination sub-unit is configured to determine the custom simplification algorithm as the target simplification algorithm if the target user performs a confirmation operation indicating that the custom simplification algorithm is applicable to the parameter trial.

[0352] As a further embodiment, the apparatus further comprises:

[0353] A parameter determination module is configured to determine at least one candidate parameter that does not satisfy a simplification condition from a plurality of candidate parameters, and obtain trial results respectively corresponding to the at least one candidate parameter.

[0354] A parameter selection module is configured to select a target parameter satisfying an optimal parameter condition from the at least one candidate parameter based on the trial results respectively corresponding to the at least one candidate parameter.

[0355] As an embodiment, the apparatus can further comprise:

[0356] A resource request module is configured to receive a parameter optimization request initiated by a to-be-processed parameter for a target resource.

[0357] The first response module can comprise:

[0358] A first response unit is configured to determine a target function corresponding to a processing target of the target resource in response to the parameter optimization request.

[0359] The apparatus can further comprise:

[0360] A first sampling module is configured to perform multiple sampling processes on the to-be-processed parameter to obtain a plurality of candidate parameters.

[0361] The resource processing module is configured to generate processing information of the target resource according to a value of the target parameter of the to-be-processed parameter, and to process the target resource according to the processing information.

[0362] As a further example, the apparatus can further include:

[0363] The browsing response module is configured to detect a browsing operation initiated by a target user, and generate a parameter optimization request for a browsing parameter of the target user.

[0364] The first response module can include:

[0365] The second response unit is configured to determine a target function corresponding to an access target of the target user in response to the parameter optimization request.

[0366] The apparatus can further include:

[0367] The second sampling module is configured to perform multiple sampling processes on the browsing parameter to obtain multiple candidate parameters.

[0368] The information generation module is configured to generate access recommendation information of the target user according to a value of the target parameter of the browsing parameter.

[0369] The product matching module is configured to search for a target product matched with the access recommendation information from a product database, and to output the target product for the target user.

[0370] Figure 9 The implementation data processing apparatus can perform Figure 1 The data processing method of the embodiments shown in the above description has the same implementation principles and technical effects as the data processing method of the embodiments shown in the above description. The specific manners of the steps performed by the processing components in the above embodiments have been described in detail in the embodiments related to the method, and will not be described in detail here.

[0371] In actual applications, Figure 10 The data processing apparatus shown in the above description can be configured as a computing device, and the computing device is described in detail in the above description. Figure 10 is a structural schematic diagram of an embodiment of a computing device provided by the embodiments of the present application. The device can include a storage component 1001 and a processing component 1002. The storage component 1001 is configured to store one or more computer instructions. The one or more computer instructions are invoked by the processing component 1002 to execute the hyperparameter optimization method shown in the embodiments such as the embodiments shown in the above description. Figure 1

[0372] ​The processing component 1002 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, for executing the above method.

[0373] The storage component 1001 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0374] Of course, the computing device can also include other components, such as an input / output interface, a communication component, etc. The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc. The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0375] In addition, the embodiments of the present application also provide a computer readable storage medium, which can store one or more computer instructions, and the one or more computer instructions are executed to implement any one of the data processing methods in the embodiments of the present application.

[0376] As shown in FIG. 10, a structure schematic diagram of another embodiment of a test simplification device provided by the embodiments of the present application is shown, which can include: Figure 11 The request receiving module 1101 is configured to receive a judgment request initiated by the computing device for whether the parameter test of any candidate parameter on the target function satisfies the simplification condition.

[0377] The target function is determined by the computing device in response to a parameter optimization request.

[0378] The second response module 1102 is configured to respond to the judgment request and judge whether the parameter test satisfies the simplification condition.

[0379] The parameter test of the candidate parameter is stopped when the simplification condition is satisfied; the parameter test of the candidate parameter is continued to obtain the test result of the candidate parameter when the simplification condition is not satisfied.

[0380]

[0381] ​As an embodiment, the apparatus further comprises:

[0382] The second determining module is configured to determine a target reduction algorithm in the plurality of trial reduction algorithms that matches the parameter trial.

[0383] The third judging module is configured to determine, according to the target reduction algorithm, whether the parameter trial satisfies the reduction condition, and generate a judgment result.

[0384] Optionally, the target reduction algorithm is determined from the plurality of trial reduction algorithms in the following manner: searching, from the plurality of trial reduction algorithms of the trial reduction module, the target reduction algorithm that matches the parameter trial.

[0385] In some embodiments, the third judging module can be specifically configured to determine, based on the target reduction algorithm, whether the intermediate trial result of the parameter trial satisfies the reduction condition, and generate a judgment result.

[0386] Figure 11 The implementation trial reduction apparatus can perform Figure 7 The data processing method of the embodiments shown in the above description has the same implementation principles and technical effects. The specific manners in which the processing components perform the steps in the above embodiments have been described in detail in the embodiments of the method, and thus will not be described in detail here.

[0387] In actual applications, Figure 11 The trial reduction apparatus shown in the above description can be configured as a trial reduction device, and the above description can be referred to. Figure 12 FIG. 1 shows a structural schematic diagram of an embodiment of a trial reduction device provided by the embodiments of the present application. The device can include a storage component 1201 and a processing component 1202. The storage component 1201 is configured to store one or more computer instructions. The one or more computer instructions are invoked by the processing component 1202 to execute the data processing method shown in the embodiments of the above description. Figure 7 The data processing method shown in the embodiments of the above description.

[0388] The processing component 1202 can include one or more processors to execute the computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components, configured to execute the above method.

[0389] The storage component 1201 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or nonvolatile storage devices or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0390] Of course, the computing device can also include other components, such as an input / output interface, a communication component, and the like. The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, and the like. The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, and the like.

[0391] In addition, the embodiments of the present application also provide a computer readable storage medium, which can store one or more computer instructions, and the one or more computer instructions are executed to implement any one of the data processing methods in the embodiments of the present application.

[0392] As shown in FIG. 13, FIG. 13 is a structural schematic diagram of another embodiment of a data processing system provided by the embodiments of the present application, which can include a computing module 1301 and a test simplification module 1302. Figure 13

[0393] The computing module is configured to: in response to a parameter optimization request, determine a target function corresponding to the parameter optimization request; in a process in which the target function performs parameter test on any candidate parameter, generate a judgment request for judging whether the parameter test meets a simplification condition; send the judgment request to the test simplification module; determine a judgment result corresponding to the judgment request; if the judgment result is that the parameter test meets the simplification condition, stop the parameter test of the candidate parameter; if the judgment result is that the parameter test does not meet the simplification condition, continue to perform the parameter test of the candidate parameter to obtain a test result of the candidate parameter.

[0394] The test simplification module is configured to: obtain the judgment request sent by the computing module; in response to the judgment request, judge whether the parameter test meets the simplification condition.

[0395] As an embodiment, the computing module, in the process in which the target function performs parameter test on any candidate parameter, generates the judgment request for judging whether the parameter test meets the simplification condition, specifically includes:

[0396] In the process in which the target function performs parameter test on any candidate parameter, the judgment request for judging whether the parameter test meets the simplification condition is generated based on a target simplification algorithm in a plurality of test simplification algorithms of the test simplification module. ​

[0397] The computing module in the embodiments of the present application can be specifically the computing device involved in the foregoing embodiments, and the test simplification module can be specifically the test simplification device in the foregoing embodiments. The specific content and technical effects of the computing model and the test simplification module have been described in detail in the foregoing embodiments, and will not be described here.

[0398] The device embodiments described above are merely schematic, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments according to actual needs. Those skilled in the art can understand and implement it without creative labor.

[0399] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of a general hardware platform as necessary, and of course can also be realized by means of combination of hardware and software. Based on such understanding, the above technical solutions can be embodied in the form of a computer product, and the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0400] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A data processing method, characterized in that, include: In response to a parameter optimization request, determine the objective function corresponding to the parameter optimization request; The objective function is divided into multiple functions according to the calculation order. The calculation stage of the multiple functions is the corresponding parameter experiment stage. A monitoring node is set at the calculation result and / or in the middle of the calculation process of any of the multiple functions. During the parameter experiment of the objective function on any candidate parameter, the objective simplification algorithm among the multiple experiment simplification algorithms of the experiment simplification module is called. When the objective function runs to any monitoring node, the intermediate test results of the candidate parameter at the monitoring node are obtained, and it is judged whether the intermediate test results meet the simplification conditions, and the judgment result of the candidate parameter is obtained. The target simplification algorithm applies to one or more of the following: node information of the test monitoring node, parameter test type, and parameter test stage, which match the parameter test. If the judgment result indicates that the parameter test meets the simplification condition, then the parameter test of the candidate parameter is stopped; If the judgment result indicates that the parameter test does not meet the simplification condition, then the parameter test of the candidate parameter is continued to be performed to obtain the test result of the candidate parameter.

2. The method according to claim 1, characterized in that, The test simplification module includes multiple test simplification algorithms; During the parameter experimentation process of the objective function for any candidate parameter, the experiment simplification module is invoked to determine whether the parameter experiment meets the simplification conditions, and the determination result is obtained as follows: During the parameter experimentation process of the objective function for any candidate parameter, the objective simplification algorithm among the multiple experiment simplification algorithms of the experiment simplification module is called to determine whether the parameter experiment meets the simplification conditions, and the determination result is obtained.

3. The method according to claim 2, characterized in that, The target simplification algorithm was determined from the plurality of experimental simplification algorithms in the following manner: The target simplification algorithm that matches the parameter experiment is found from among the multiple simplification algorithms in the simplification module.

4. The method according to claim 3, characterized in that, The step of finding the target simplification algorithm that matches the parameter experiment from multiple test simplification algorithms in the test simplification module includes: Determine the first test information corresponding to the parameter test of the candidate parameters; Obtain the second test information associated with the multiple test simplification algorithms respectively; Query the target test information that matches the first test information from the second test information corresponding to the multiple test simplification algorithms; The test simplification algorithm corresponding to the target test information is determined to be the target simplification algorithm.

5. The method according to claim 4, characterized in that, The first test information includes: parameter test type; The query for target test information that matches the first test information in the second test information corresponding to the multiple test simplification algorithms includes: From the second test information corresponding to the multiple test simplification algorithms, query the target test information that matches the parameter test type of the first test information.

6. The method according to claim 5, characterized in that, The parameter test types include: serial test type and parallel test type; The step of querying target test information that matches the parameter test type of the first test information from the second test information corresponding to the plurality of test simplification algorithms includes: If the parameter test type of the candidate parameter is the serial test type, determine the target test information containing the serial test type from the second test information corresponding to the multiple test simplification algorithms; or, If the parameter test type of the candidate parameter is the parallel test type, the target test information containing the parallel test type is determined from the second test information corresponding to the multiple test simplification algorithms.

7. The method according to claim 4, characterized in that, The first test information includes: the parameter test phase; The query for target test information that matches the first test information in the second test information corresponding to the multiple test simplification algorithms includes: From the second test information corresponding to the multiple test simplification algorithms, query the target test information that matches the parameter test stage of the first test information.

8. The method according to claim 4, characterized in that, The first test information includes: the parameter test stage and the parameter test type; The query for target test information that matches the first test information in the second test information corresponding to the multiple test simplification algorithms includes: Target test information that simultaneously matches the parameter test type and the parameter test stage is determined from the second test information corresponding to the multiple test simplification algorithms.

9. The method according to claim 2, characterized in that, The multiple trial simplification algorithms include a custom simplification algorithm set by the target user; The target simplification algorithm was also determined from the plurality of experimental simplification algorithms in the following manner: If among the multiple experimental simplification algorithms there is a custom simplification algorithm set by the target user, then the custom simplification algorithm is determined to be the target simplification algorithm.

10. The method according to claim 9, characterized in that, Also includes: Based on the target user's custom simplification algorithm, the test simplification module is controlled to save the custom simplification algorithm.

11. The method according to claim 9, characterized in that, If among the plurality of experimental simplification algorithms there exists a custom simplification algorithm set by the target user, then determining the custom simplification algorithm as the target simplification algorithm includes: If the target user's custom simplification algorithm exists among the multiple test simplification algorithms, a prompt message indicating the existence of the custom simplification algorithm is generated; The prompt message is displayed to the target user so that the target user can confirm whether the custom algorithm is suitable for the parameter experiment; If the target user performs a confirmation operation for the applicability of the custom simplification algorithm to the parameter test, the custom simplification algorithm is determined to be the target simplification algorithm.

12. The method according to claim 1, characterized in that, The target simplification algorithm includes a historical estimation algorithm; The historical estimation algorithm specifically determines whether the intermediate experimental results meet the simplification criteria in the following ways: Based on the historical intermediate results and historical experimental results corresponding to multiple historical parameters, the estimated experimental results corresponding to the experimental intermediate results are predicted. Determine whether the estimated test result matches the result threshold; If not, then the intermediate results of the experiment are determined to meet the simplification conditions; If so, then the intermediate results of the experiment do not meet the simplification conditions.

13. The method according to claim 3, characterized in that, The target simplification algorithm includes: a calculation comparison algorithm; The calculation and comparison algorithm specifically determines whether the intermediate experimental results meet the simplification criteria in the following ways: During the experiment, the intermediate reference values ​​corresponding to the monitoring nodes that obtain the intermediate results of the experiment are determined. Determine whether the intermediate reference value meets the preset reference threshold; If not, then the intermediate results of the experiment are determined to meet the simplification conditions; If so, then the intermediate results of the experiment do not meet the simplification conditions.

14. The method according to claim 1, characterized in that, Also includes: Identify at least one candidate parameter among multiple candidate parameters that does not meet the simplification conditions, and obtain the experimental results corresponding to the at least one candidate parameter respectively; Based on the experimental results corresponding to the at least one candidate parameter, a target parameter that satisfies the optimal parameter condition is selected from the at least one candidate parameter.

15. The method according to claim 14, characterized in that, Also includes: Receive parameter optimization requests for pending parameters of the target resource; The step of determining the objective function corresponding to the parameter optimization request in response to the parameter optimization request includes: In response to the parameter optimization request, determine the objective function corresponding to the processing target of the target resource; Also includes: The parameters to be processed are sampled multiple times to obtain multiple candidate parameters; After selecting the target parameter that satisfies the parameter optimality condition from the at least one candidate parameter based on the test results corresponding to the at least one candidate parameter, the method further includes: Based on the value of the target parameter of the parameter to be processed, processing information of the target resource is generated, and the target resource is processed according to the processing information.

16. The method according to claim 14, characterized in that, Also includes: Detect browsing operations initiated by the target user and generate a parameter optimization request for the browsing parameters of the target user; The step of determining the objective function corresponding to the parameter optimization request in response to the parameter optimization request includes: In response to the parameter optimization request, determine the objective function corresponding to the access target of the target user; Also includes: The browsing parameters are sampled multiple times to obtain multiple candidate parameters; After selecting the target parameter that satisfies the parameter optimality condition from the at least one candidate parameter based on the test results corresponding to the at least one candidate parameter, the method further includes: Based on the values ​​of the target parameters in the browsing parameters, generate access recommendation information for the target user; The system searches the product database for target products that match the access recommendation information and outputs the target products to the target user.

17. A data processing method, characterized in that, include: In response to a request to call the parameter processing interface, determine the processing resources corresponding to the parameter processing interface; The following steps are performed using the processing resources corresponding to the parameter processing interface: In response to a parameter optimization request, the objective function corresponding to the parameter optimization request is determined; the objective function is divided into multiple functions according to the calculation order, and the calculation stage of the multiple functions is the corresponding parameter experiment stage; a monitoring node is set at the calculation result and / or in the middle of the calculation process of any of the multiple functions. During the parameter experimentation process of the objective function for any candidate parameter, the objective simplification algorithm among the multiple experiment simplification algorithms of the experiment simplification module is called. When the objective function reaches any monitoring node, the intermediate experimental results of the candidate parameter at that monitoring node are obtained, and it is judged whether the intermediate experimental results meet the simplification conditions, and the judgment result is obtained. The node information, parameter experiment type, and parameter experiment stage of the experimental monitoring node to which the objective simplification algorithm is applicable are matched with the parameter experiment. If the judgment result indicates that the parameter test meets the simplification condition, then the parameter test of the candidate parameter is stopped; If the judgment result indicates that the parameter test does not meet the simplification condition, then the parameter test of the candidate parameter is continued to be performed to obtain the test result of the candidate parameter.

18. A data processing method, characterized in that, include: The system receives a judgment request from the computing device regarding whether the intermediate test results of the parameter experiment on the candidate parameters for the objective function meet the simplification conditions; wherein, the objective function is determined by the computing device in response to the parameter optimization request; the objective function is divided into multiple functions according to the calculation order, the calculation stage of the multiple functions is the corresponding parameter experiment stage, and a monitoring node is set at the calculation result and / or the middle of the calculation process of any of the multiple functions; In response to the judgment request, the target simplification algorithm among multiple test simplification algorithms is used to obtain the intermediate test results of the candidate parameters at any monitoring node when the objective function runs to any monitoring node, and to determine whether the intermediate test results meet the simplification conditions; the node information, parameter test type, and parameter test stage of the test monitoring node to which the target simplification algorithm is applicable are one or more of the following, and they match the parameter test. The parameter experiment of the candidate parameter is stopped when the simplification condition is met; the parameter experiment of the candidate parameter continues to be executed to obtain the test results of the candidate parameter when the simplification condition is not met.

19. A data processing apparatus, characterized in that, include: The first response module is used to respond to the parameter optimization request and determine the objective function corresponding to the parameter optimization request; The objective function is divided into multiple functions according to the calculation order. The calculation stage of the multiple functions is the corresponding parameter experiment stage. A monitoring node is set at the calculation result and / or in the middle of the calculation process of any of the multiple functions. The result acquisition module is used to call the target simplification algorithm among multiple test simplification algorithms in the test simplification module during the parameter experiment of the objective function for any candidate parameter. When the objective function runs to any monitoring node, it acquires the intermediate test results of the candidate parameter at that monitoring node, judges whether the intermediate test results meet the simplification conditions, and obtains the judgment result. The node information, parameter test type, and parameter test stage of the test monitoring node to which the target simplification algorithm is applicable are matched with the parameter test. The first processing module is used to stop the parameter test of the candidate parameter if the judgment result is that the parameter test meets the simplification condition. The second processing module is used to continue executing the parameter test of the candidate parameter if the judgment result is that the parameter test does not meet the simplification condition, and to obtain the test result of the candidate parameter.

20. A data processing apparatus, characterized in that, include: The request receiving module is used to receive a judgment request initiated by the computing device regarding whether the intermediate test results of the parameter experiment of any candidate parameter for the objective function meet the simplification conditions; wherein, the objective function is determined by the computing device in response to the parameter optimization request; the objective function is divided into multiple functions according to the calculation order, the calculation stage of the multiple functions is the corresponding parameter experiment stage, and a monitoring node is set at the calculation result and / or the middle of the calculation process of any of the multiple functions; The second response module is used to respond to the judgment request, and utilize the target simplification algorithm among multiple test simplification algorithms to obtain the intermediate test results of the candidate parameters at any monitoring node when the objective function runs to any monitoring node, and determine whether the intermediate test results meet the simplification conditions; the node information, parameter test type, and parameter test stage of the test monitoring node to which the target simplification algorithm is applicable are one or more of the following, and are matched with the parameter test; The parameter experiment of the candidate parameter is stopped when the simplification condition is met; the parameter experiment of the candidate parameter continues to be executed to obtain the test results of the candidate parameter when the simplification condition is not met.

21. A computing device, characterized in that, It includes: a storage component and a processing component; the storage component is used to store one or more computer instructions; the one or more computer instructions are invoked by the processing component to execute the method according to any one of claims 1 to 16.

22. A simplified testing device, characterized in that, include: A storage component and a processing component; the storage component is used to store one or more computer instructions; the one or more computer instructions are invoked by the processing component to execute the method of claim 18.

Images