WO2022156731A1 - 信息处理方法及装置、服务器及用户设备 - Google Patents

信息处理方法及装置、服务器及用户设备 Download PDF

Info

Publication number
WO2022156731A1
WO2022156731A1 PCT/CN2022/072913 CN2022072913W WO2022156731A1 WO 2022156731 A1 WO2022156731 A1 WO 2022156731A1 CN 2022072913 W CN2022072913 W CN 2022072913W WO 2022156731 A1 WO2022156731 A1 WO 2022156731A1
Authority
WO
WIPO (PCT)
Prior art keywords
target user
test sample
parameter
sampling algorithm
simulation result
Prior art date
Application number
PCT/CN2022/072913
Other languages
English (en)
French (fr)
Inventor
邹晨俊
Original Assignee
阿里巴巴集团控股有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 阿里巴巴集团控股有限公司 filed Critical 阿里巴巴集团控股有限公司
Priority to EP22742210.2A priority Critical patent/EP4283511A4/en
Publication of WO2022156731A1 publication Critical patent/WO2022156731A1/zh
Priority to US18/357,442 priority patent/US20240020556A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/15—Correlation function computation including computation of convolution operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0242—Determining effectiveness of advertisements
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0251—Targeted advertisements
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/06—Power analysis or power optimisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn

Definitions

  • the present application relates to the technical field of electronic devices, and in particular, to an information processing method and apparatus, a server, and user equipment.
  • the black-box algorithm is an algorithm model established to solve complex problems that cannot be modeled using explicit mathematical formulas.
  • the mathematical model of the black-box algorithm is usually unknown, and the corresponding input data is input into the black-box algorithm.
  • the objective function of calculate the output data corresponding to the input data.
  • the specific objective function of the black-box algorithm is unknown. Due to the high service types and selectivity of black-box algorithms, black-box algorithms can be applied to many fields such as e-commerce, finance, and electricity.
  • black-box optimization based on the AutoML (Automated machine learning, automatic machine learning) framework.
  • the main process of black-box optimization is to continuously generate new samples through the parameter sampling algorithm for the parameters to be optimized, and then use the objective function of the black box to evaluate the use effect of the samples to obtain the simulation results of the samples.
  • the algorithm generates new samples and performs simulation evaluation to generate target samples that are close to the simulation results and meet the predetermined conditions as soon as possible.
  • the target samples and the simulation results corresponding to the target samples are the optimization results of this black-box optimization.
  • the black-box optimization process is mainly performed locally, resulting in low efficiency of the black-box optimization.
  • the embodiments of the present application provide an information processing method and apparatus, a server, and a user equipment, so as to solve the technical problem in the prior art that the black box optimization process is performed locally, resulting in low efficiency of black box optimization.
  • an embodiment of the present application provides an information processing method, including:
  • the simulation result of the test sample is output for the target user.
  • the embodiments of the present application provide an information processing method, including:
  • the simulation result of the test sample is output for the target user.
  • the embodiments of the present application provide an information processing method, including:
  • the simulation result of the test sample is output for the target user.
  • an information processing apparatus including:
  • a request detection module used to detect a parameter optimization request initiated by a target user, and determine a parameter sampling algorithm that matches the target user
  • a first response module configured to call the parameter sampling algorithm to generate a test sample in response to a sample acquisition request initiated by the target user
  • a first simulation module configured to determine the simulation result of the test sample based on a preset objective function
  • the first output module is used for outputting the simulation result of the test sample for the target user.
  • an information processing apparatus including:
  • a first sending module configured to send a parameter optimization request provided by a target user to a server, so that the server can determine a parameter sampling algorithm that matches the target user;
  • a second sending module configured to send a sample acquisition request initiated by the target user to the server, so that the server responds to the sample acquisition request and invokes the parameter sampling algorithm to generate a test sample
  • a second simulation module configured to determine the simulation result of the test sample based on a preset objective function
  • the second output module is configured to output the simulation result of the test sample for the target user.
  • an embodiment of the present application provides a server, including: 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;
  • the processing components are used to:
  • an embodiment of the present application provides a user terminal, including: 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 called by the processing component ;
  • the processing components are used to:
  • Detect the parameter optimization request initiated by the target user for the parameter sampling algorithm send the parameter optimization request provided by the target user to the server, so that the server can determine the parameter sampling algorithm that matches the target user; send the target user to initiate
  • the sample acquisition request is sent to the server, so that the server responds to the sample acquisition request and invokes the parameter sampling algorithm to generate a test sample; based on the preset objective function, determine the simulation result of the test sample;
  • the target user outputs the simulation result of the test sample.
  • a parameter optimization request initiated by a target user may be received, and a parameter sampling algorithm matching the target user may be determined. Then, in response to the sample acquisition request initiated by the target user, the parameter sampling algorithm can be invoked to generate the test sample.
  • the test sample can be used to determine the simulation result in the calculation method of the objective function, so as to output the simulation result of the test sample for the target user.
  • the user's service needs can be obtained, and by effectively responding and providing feedback to the user's service needs, it can effectively provide users with black-box services and realize the extended application of black-box services. Improve the utilization efficiency of black-box algorithms.
  • FIG. 1 is a flowchart of an embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 2 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 3 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 4 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 5 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 6 is a flowchart of another embodiment of an information control method provided by an embodiment of the present application.
  • FIG. 7 is an example application diagram of an information processing method provided by an embodiment of the present application.
  • FIG. 8 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 9 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of an embodiment of an information processing apparatus provided by an embodiment of the application.
  • FIG. 11 is a schematic structural diagram of an embodiment of a server according to an embodiment of the present application.
  • FIG. 12 is a schematic structural diagram of another embodiment of an information processing apparatus provided by an embodiment of the present application.
  • FIG. 13 is a schematic structural diagram of an embodiment of a user equipment according to an embodiment of the present application.
  • the words “if”, “if” as used herein may be interpreted as “at” or “when” or “in response to determining” or “in response to identifying”.
  • the phrases “if it is determined” or “if a (stated condition or event) is identified” can be interpreted as “when determined” or “in response to a determination” or “when a (stated condition or event) is identified,” depending on the context )” or “in response to an identification (statement or event)”.
  • the technical solutions of the embodiments of the present application can be applied in the process of parameter optimization.
  • the multi-user application of black box optimization can be realized, the scope of application of black box optimization can be expanded, and the utilization efficiency of black box optimization can be improved. .
  • black-box optimization service can be used.
  • the main process of black-box optimization is to use the black-box algorithm to continuously generate new samples, and use the simulation function that simulates the actual calculation effect of the user, that is, the objective function to simulate the use effect of the generated new sample, and obtain the simulation result of the sample.
  • the effects of the newly generated samples are continuously simulated, so as to select the sample with the highest use effect from the obtained samples as the global target solution.
  • the existing black-box algorithms or black-box optimization strategies are mostly used internally by developers, resulting in a narrow application range of black-box optimization and low utilization of black-box algorithms.
  • a parameter optimization request initiated by a target user may be detected, and a parameter sampling algorithm matching the target user may be determined.
  • target users can use the black box algorithm to optimize parameters at any time.
  • the simulation result of the test sample can be determined based on the preset objective function, the test sample can be automatically generated for the target user, and the simulation calculation of the test sample can be completed.
  • the simulation result is usually the estimation result of the use effect of the test sample, so as to accurately evaluate the test sample, so that after the simulation result of the test sample is output for the target user, it can be viewed by the target user and determine whether the test sample is available.
  • the effect test of the test sample Through the interaction with target users, the cloud service of parameter optimization is realized, so as to realize rapid parameter optimization in the cloud and improve the efficiency of parameter optimization.
  • the extended application of black-box services is realized, and the utilization efficiency of black-box algorithms is improved.
  • the flow chart of an embodiment of a kind of information processing method provided by the embodiment of the application may include the following steps:
  • the information processing method provided by the embodiments of the present application can be applied to the server, where the server is relative to the client, and the server and the client correspond to a C/S (Client-Server) architecture.
  • the server can be, for example, a computer, a common server, a cloud server, a super personal computer, a notebook computer, a tablet computer, or the like.
  • the client can be configured on the user equipment, and the client can be, for example, a mobile phone, a tablet computer, a computer, a virtual reality device, an augmented reality device, a wearable device, and the like.
  • the embodiments of the present application do not limit the specific types of the server and the client too much.
  • the server may be configured with multiple parameter sampling algorithms. Different parameter sampling algorithms have different principles, but the results are the same, and can be used to generate test samples.
  • the general parameter sampling algorithm can be used to generate samples of parameters to be optimized, and use a preset objective function to simulate the use effect of the sample, so as to estimate the use effect of the sample and obtain the simulation result of the sample.
  • the target samples with the highest evaluation of the simulation results can be screened, and the target samples can be output as the result of parameter optimization to complete the parameter optimization.
  • the target user can send a parameter optimization request to the server through the client.
  • the server can receive the parameter optimization request sent by the client of the target user.
  • the parameter optimization request may be generated by the target user based on the parameters to be optimized.
  • the content recommendation in the field of e-commerce as an example, when the target user browses the e-commerce website, the website can push the content that the target user is interested in for the target user. The contribution of each category is different. Therefore, the ratio of each category of recommended content can be used as the parameter to be optimized for parameter optimization.
  • the server can be configured with multiple candidate sampling algorithms, and the target user can specify the parameter sampling algorithm in the multiple black-box algorithms requested by the parameter optimization.
  • the parameter optimization request may include algorithm identification information of the parameter sampling algorithm.
  • the algorithm identification information may be, for example, an algorithm name.
  • the parameter optimization request may also include parameter information such as the sample type, value range, and quantity corresponding to the sample of the parameter to be optimized.
  • the parameter sampling algorithm can generate multiple samples of the parameters to be optimized at the same time, so as to evaluate the multiple samples at the same time and improve the efficiency of parameter optimization.
  • the parameter optimization request may further include a test identifier set for this parameter optimization test of the target user, so as to distinguish it from the optimization process of different parameters.
  • the target user can also log in to the server with user information such as user name and password to obtain the permission to use the parameter optimization service and achieve effective management of users.
  • the server when the server receives the parameter optimization request, it can store the parameter information, user information, test identification and other information in the parameter optimization request, so as to facilitate the accurate process and results of the parameter optimization service. Tracking and improving the efficiency of information processing.
  • the server After the server determines the parameter sampling algorithm that matches the target user, it can feed back the startup prompt information of startup parameter optimization to the target user. After the target user sends the parameter optimization request, it can receive the start prompt message corresponding to the parameter optimization fed back by the server.
  • a sample acquisition request can be sent to the server. The server can receive the sample acquisition request sent by the target user, and in response to the sample acquisition request, call the parameter sampling algorithm to generate test samples.
  • the test sample can be a parameter value set for the parameter to be optimized, and specifically can be a candidate solution generated by the parameter sampling algorithm for the parameter to be optimized.
  • the candidate solution satisfies the optimization conditions, it constitutes the final optimization result of the parameter to be optimized.
  • the objective function can simulate the use effect or use result of the test sample to obtain the simulation result.
  • the objective function can be determined by the optimization objective of the parameter to be optimized.
  • the parameter to be optimized can be a parameter corresponding to the recommended content
  • the test sample is the recommended content found to the browsing user based on the parameter to be optimized.
  • the prediction model of the click-through rate can be used as the objective function to predict the click-through rate of the browsing users on the recommended content, and the predicted click-through rate is the Simulation results of the test sample.
  • outputting the simulation result of the test sample for the target user may include: sending the simulation result of the test sample to the user terminal of the target user, so that the user terminal can display the output result of the test sample.
  • the server when the server sends the simulation result of the test sample to the client of the target user, it can be sent to the client of the target user in the form of a web page, instant messaging message, short message, etc.
  • a parameter optimization request initiated by a target user may be detected, and a parameter sampling algorithm matching the target user may be determined.
  • target users can use the black box algorithm to optimize parameters at any time.
  • the simulation result of the test sample can be determined based on the preset objective function, the test sample can be automatically generated for the target user, and the simulation calculation of the test sample can be completed.
  • the cloud service of parameter optimization is realized, so as to realize rapid parameter optimization in the cloud and improve the efficiency of parameter optimization.
  • the simulation result of the test sample is usually an evaluation result obtained by evaluating the use effect of the test sample based on the objective function.
  • the simulation result can accurately evaluate the use effect of the test sample, so as to output the test for the target user.
  • the target user checks the simulation result of the test sample, so as to directly determine whether the test sample meets the usage requirements through the simulation result, so as to realize an accurate indication of the use effect of the test sample.
  • the CTR calculation model when the CTR calculation model is used as the objective function, the higher the calculated CTR, the more effective the recommended content is.
  • the target user can confirm The test sample meets the conditions of use.
  • the test sample can be provided to the recommender system, so that the recommender system generates recommended content based on the test sample and recommends it to users who use the recommender system.
  • the present application can be opened to many types of target users, such as system developers, scholar, students, system designers, and enterprise users, so as to provide parameter optimization services to different types of target users.
  • target users such as system developers, scholar, students, system designers, and enterprise users
  • the parameter optimization service is effectively provided for the target user
  • the target user can continuously initiate sample acquisition requests, and the server can continuously respond to the sample acquisition request, providing test samples for the target user, so that the target user can know the simulation results of each test sample, and choose the one that meets the requirements.
  • Use the conditional target sample Use the conditional target sample.
  • Provide parameter optimization services to different target users realize the extended application of parameter optimization services, allow more users to use parameter optimization services, and improve the utilization efficiency of parameter optimization.
  • the objective function may specifically belong to a black-box algorithm, which cannot be expressed by an explicit mathematical formula, and the specific mathematical model of the objective function is usually unknown.
  • the parameter optimization request may include a black-box optimization request.
  • detecting a parameter optimization request initiated by a target user, and determining a parameter sampling algorithm that matches the target user may include:
  • the technical solutions of the embodiments of the present application may be applied to a cloud server, and the cloud server may include a public cloud and a private cloud.
  • the public cloud can provide a cloud that can be used by cloud users on a third-party service. Generally, it can be used through a network to realize shared resource services, and can provide services for an open public network. Private clouds are built for single use by one customer, thus providing the most effective control over data, security, and quality of service. Therefore, the security of private cloud is higher than that of public cloud. For the sake of data security, when users use the information processing method configured in the cloud, the processing flow of public cloud and private cloud can be different.
  • the objective function When using the objective function to simulate the test sample, it involves a lot of user information or data.
  • the simulation process of the test sample can be completed in the private cloud to reduce the processing pressure on the user side.
  • the simulation process of the test sample can be completed on the user side .
  • the public cloud and the private cloud can be implemented based on the underlying services of K8S.
  • K8S Kernetes, container cluster management system
  • a cluster is a group of nodes.
  • a cloud server whether it is a public cloud or a private cloud, can be regarded as a service cluster.
  • the service cluster can be composed of multiple nodes.
  • the K8S platform is installed on these nodes, so that the K8S can be used for the services in the cluster. Nodes are managed.
  • the information processing method in the embodiment of the present application can be applied to the K8S platform, and specifically can be configured on the Custom resource controller (CRD, custom resource controller) created in the K8S platform, that is, the CRD in the K8S can be implemented as a processing component Information processing method.
  • CRD custom resource controller
  • the technical solutions of the embodiments of the present application can be configured on the basis of the K8S underlying services, the functions of each step can be modularized, and then configured on the nodes, and based on the function modules, an interface for external services can be provided.
  • the interface can realize data or information interaction with the client, and provide users with parameter optimization services.
  • the cloud server can be configured with gateways, data storage components, black box algorithm components, early stop components, web (World Wide Web, global wide area network) pages, as well as sample collection systems, databases and other components. The cooperation of the components completes the technical solutions of the embodiments of the present application.
  • the gateway may be the Rest (Representational State Transfer) API (Application Programming Interface) gateway API Server (application programming interface service) of K8S;
  • the data storage component may be The distributed storage component Etcd of K8S (stores cluster status information);
  • the black-box algorithm component can provide developers with the core algorithm component Suggestion Service (suggestion service), for example, it can include: random search (random search)/grid search (hyperparameter One or more of optimization) / cmaes (maximum likelihood estimation) / pma (Partitioning Around Medoid, division around the center point); early stop (Early Stop) component can be used to stop the process is not ideal through early stop technology
  • the parameter simulation of the test sample can be stopped early, reducing resource consumption and accelerating the test progress;
  • the web page is mainly used to output test samples and test sample simulation results to external users, so that users can view them in time;
  • the sample collection system Metal System
  • the simulation process of the test sample can be completed on the client side to reduce the possibility of data leakage and improve data security.
  • FIG. 2 which is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application, the method may include the following steps:
  • 201 Detect a parameter optimization request initiated by a target user, and determine a parameter sampling algorithm matching the target user.
  • the parameter optimization request may be sent by the client to the server, and the server may provide an interface for external communication, that is, an information processing interface, and receive the parameter optimization request sent by the client through the information processing interface.
  • the requested service information which can be formed by information such as algorithm information, parameter information, user information and/or verification information input on the client, can be displayed in the form of SDK (Software Development Kit). , software development kit) to the server, the information processing interface of the server can receive the parameter optimization request encapsulated in the form of SDK, and obtain the parameter sampling algorithm corresponding to the requested service information.
  • SDK Software Development Kit
  • the method may further include: receiving a sample test request sent by the client of the target user, and in response to the sample test request, sending the test sample to the client of the target user.
  • test prompts can be given to target users, for example, test prompt information can be generated, and the test prompt information can include risk prompt information for sample testing in the public cloud, for example, in the form of text messages.
  • the target user "is there a risk of data leakage if the simulation calculation of the test sample is performed on the public cloud".
  • the user terminal can output the risk prompt information for the target user, so that the target user can view the risk prompt information and make corresponding feedback operations on the risk prompt information.
  • the target user can directly perform a confirmation operation by simulating the test sample on the client side, and when the server side detects the confirmation operation, it can send the test sample to the client side of the target user.
  • outputting the simulation result of the test sample for the target user may include: controlling the target user to output the simulation result of the test sample, for example, when the server receives the simulation result of the test sample, the simulation result may be verified, and after the verification is successful, The output control information of the simulation result can be output to the user terminal, and the user terminal can respond to the output control information of the simulation result and output the simulation result.
  • the parameter optimization request initiated by the target user can be detected, the parameter sampling algorithm matching the target user can be determined, and the parameter sampling algorithm can be invoked to generate the test sample in response to the sample acquisition request initiated by the target user.
  • the test sample is sent to the user terminal of the target user, so that the user terminal can input the test sample into the preset objective function, and obtain the simulation result corresponding to the test sample by calculation.
  • the test work of the test sample is placed on the user side, which reduces data transmission and improves interaction efficiency. After receiving the simulation result corresponding to the test sample sent by the user terminal, the simulation result of the test sample can be output for the target user.
  • FIG. 3 is a flowchart of another embodiment of an information processing method provided by an embodiment of the present application, the method may include the following steps:
  • the parameter optimization request may be sent by the client to the server, and the server may provide an interface for external communication, that is, an information processing interface, and receive the parameter optimization request sent by the client through the information processing interface.
  • the requested service information which may be formed by the algorithm information, parameter information, user information and/or verification information input at the client, may be in the form of a declarative API (Application Programming Interface, application program interface) is sent to the server.
  • Declarative means that the client submits an API object defined for the requested service information to declare the specific content and type of the desired requested service information.
  • the declarative API can contribute to the project orchestration capability of K8S services, so that parameter optimization requests can interact between the client and the server according to a certain object format to achieve accurate information transmission.
  • the server When the server receives the parameter optimization request, it can parse the information according to the information type and quantity defined in the parameter optimization request to obtain the relevant information in the parameter optimization request, so as to determine the parameter sampling algorithm that matches the target user.
  • the black box algorithm component may actually include multiple classical algorithms, and the parameter sampling algorithm is one of them.
  • the parameter sampling algorithm may be directly specified by the target user, that is, this , the parameter optimization request may include algorithm identification information of the parameter sampling algorithm.
  • the parameter sampling algorithm may be determined by the server according to the parameter information provided by the target user, so as to select a more matching black box algorithm for the parameter information, thereby improving the applicability of the parameter sampling algorithm to the current parameter optimization requirements.
  • inputting the test sample into a preset objective function, and calculating and obtaining the simulation result of the test sample may include: determining a scheduling execution unit (Pod) of the target function, inputting the test sample into the target function in the scheduling execution unit, and by The scheduling execution unit calculates and obtains the simulation result of the test sample.
  • the scheduling execution unit is the computing module provided by the K8S service.
  • the server when the server detects a parameter optimization request initiated by a target user and determines a parameter sampling algorithm that matches the target user, it can call the parameter sampling algorithm to generate a test sample in response to a sample acquisition request initiated by the target user.
  • the test sample is input into the preset objective function, the simulation result of the test sample is obtained by calculation, and the simulation result of the test sample is output for the target user.
  • the simulation process of the test sample is completed on the server side, which can reduce the processing pressure on the user side.
  • the computing efficiency of the server side is generally higher than that of the user side, the processing efficiency of the test sample can be improved.
  • target users can use the black box algorithm to optimize parameters at any time, and improve the utilization efficiency of the black box algorithm.
  • the cloud server can be obtained by integrating multiple nodes, and the parameter sampling algorithm can be run by one of the nodes.
  • the method further includes:
  • invoking the parameter sampling algorithm to generate the test sample includes:
  • the cloud server is based on a large-scale server cluster, which may be composed of multiple computing nodes.
  • the target node in the startup parameter sampling algorithm can be determined through node query.
  • determining the target node for starting the parameter sampling algorithm may include: querying the target node for starting the black-box algorithm from a plurality of nodes through a reverse proxy.
  • the running state of the black box algorithm in each node can be detected.
  • Black box algorithms can be configured in different nodes, and one or more black box algorithms can be configured in any node.
  • the parameter sampling algorithm can run on a node in the cloud server, and the client can actually obtain test samples from the target node running the parameter sampling algorithm.
  • the communication between the client and the target node can be realized through a router.
  • input the test sample into a preset objective function, and calculate and obtain the simulation result of the test sample may include:
  • the objective function needs to rely on the support of the calculation module during the running process. Therefore, in order to enable the objective function to simulate the test samples normally, a simulation module is established for the objective function.
  • the simulation module can be a Pod.
  • the server can receive parameter optimization requests sent by multiple clients at the same time, so as to serve multiple clients at the same time, or one client can create different parameter optimization requests for different optimization requirements, and the server can also receive one client at the same time. Multiple parameter optimization requests are sent. Therefore, in order to distinguish different black box optimization processes, the target user can send a simulation test identifier to the server at the same time when initiating a parameter optimization request to distinguish different black box services.
  • the method may also include:
  • the simulation result of the test sample is stored for the target user.
  • receiving the simulation test identifier sent by the target user may specifically be receiving the simulation test identifier sent by the target user when detecting a parameter optimization request initiated by the target user.
  • the simulation experiment identifier is used to identify the simulation experiment of the test sample in the parameter optimization request initiated by the target user. For example, when target user A and target user B initiate a parameter optimization request at the same time, when target user A initiates a parameter optimization request, they can also send a simulation test identifier a to the cloud server, and when target user B initiates a parameter optimization request, they can also send a simulation Test ID b to the cloud server.
  • the test sample generated for the target user A and the simulation result of the test sample can be identified by the simulation test identifier a.
  • the target user B initiates a sample acquisition request
  • the test sample generated for the target user B and the simulation result of the test sample can be identified by the simulation test identification b, so as to be different from the test sample identified by the simulation test identification a generated for the target user A and simulation results.
  • the black-box optimization process can be started based on the parameter sampling algorithm specified by the user.
  • FIG. 4 a flowchart of another embodiment of an information processing method provided by an embodiment of the present application, the method may include:
  • the parameter optimization request includes request service information for the parameter sampling algorithm.
  • the method may further include: sending a plurality of black box algorithms to the user terminal of the target user, so that the user terminal can output a plurality of black box algorithms for the target user, and obtain the algorithm from the target user.
  • Parameter sampling algorithm is selected among multiple black-box algorithms.
  • the requested service information may include algorithm name, algorithm identification, and other information that can be used to identify the black box algorithm.
  • Request service information composed of algorithm information, parameter information, user information and/or verification information entered by the user terminal.
  • the server After the server performs operations such as verification on the parameter optimization request, it can store the request service information.
  • K8S services, etcd distributed consensus protocol
  • Etcd is a distributed storage component developed in kubenetes, which can store network configuration and state information of objects in server clusters.
  • the storage operation of etcd requesting service information may specifically be detecting the storage action of etcd.
  • it may initiate a storage message to the processing component, so that the storage component can timely determine that the data storage component has a storage action requesting service information.
  • the parameter optimization request initiated by the target user may include request service information of the parameter sampling algorithm specified by the target user.
  • the server can detect the storage behavior of the data storage component, and when detecting that the data storage component has a storage operation requesting service information, it can determine the parameter sampling algorithm for the target user to request the service in the requesting service information.
  • the parameter sampling algorithm may be invoked to generate a test sample in response to a sample acquisition request initiated by the target user. Based on the preset objective function, the simulation result of the test sample is determined, so as to output the simulation result of the test sample for the target user.
  • target users can choose parameter sampling algorithms for parameter optimization at any time, and improve the utilization efficiency of black box algorithms.
  • the request service information may also include verification information.
  • the verification information may include identity verification information, authority verification information, and the like of the target user.
  • the method may further include:
  • the requested service information is stored in the data storage component.
  • the verification information may include the identity verification information and authority verification information of the target user, and based on the verification information in the requested service information, the parameter optimization request of the target user is verified, and obtaining the verification result may include: Request the identity verification information in the service information to verify the user identity of the target user, and obtain the identity verification result;
  • the use authority of the target user is verified based on the authority verification information in the request service information, and the authority verification result is obtained.
  • determining the verification result may include: if the identity verification result is that the verification is successful and the authority verification result is that the verification is successful, then determining that the verification result is the verification result. The test was successful. If the result of identity verification is verification failure or the result of authority verification is verification failure, it is determined that the verification result is verification failure.
  • the prompt information of the verification failure is fed back to the user terminal of the target user, so that the user terminal can output the prompt information of the verification failure for the target user to remind the user to identify Information or permission information failure reasons, and modify the information or permissions according to the failure reasons, and re-apply for the service.
  • the parameter sampling algorithm for determining that the target user requests the service in the request service information may include:
  • the parameter sampling algorithm for the target user to request the service in the request service information is determined.
  • the resource controller can be configured in the server, and the storage notification message initiated by the data storage component when storing the request service information can be obtained through the resource controller. By storing the notification message, the parameter sampling algorithm of the target user requesting the service in the requesting service information can be determined.
  • the method in response to a sample acquisition request initiated by a target user, before calling a parameter sampling algorithm to generate a test sample, the method may further include:
  • the method may further include:
  • the target sample is determined based on the test sample.
  • the process returns to responding to the sample acquisition request initiated by the target user, and the parameter sampling algorithm is invoked to generate the test sample and continue to execute.
  • determining the target sample includes:
  • the simulation result may specifically be a result index data.
  • whether the test sample satisfies the preset convergence condition can be determined by whether the number of iterations corresponding to the test sample reaches the preset number of iterations threshold. If so, it is determined that the test sample meets the preset convergence condition; if not, then Determines that the test sample does not meet the preset convergence conditions.
  • whether the test sample satisfies the preset convergence condition can be determined by whether the simulation result of the test sample satisfies the preset result threshold, if yes, it is determined that the test sample meets the preset convergence condition, if not, it is determined that the test sample The preset convergence conditions have not been met. Whether the simulation result satisfies the preset result threshold may specifically include that the simulation result is greater than the preset result threshold, or the simulation result is smaller than the preset result threshold. Whether the simulation result is larger than the result threshold or smaller than the result threshold can be specifically determined according to the calculation meaning of the objective function.
  • test samples can be generated at one time in batch mode.
  • invoking a parameter sampling algorithm to generate a test sample may include:
  • the parameter sampling algorithm is invoked in batch mode to generate multiple test samples.
  • determining the simulation result of the test sample may include:
  • the simulation results corresponding to the multiple test samples respectively are determined.
  • Simulation results for outputting test samples for target users can include:
  • the public cloud can send the multiple test samples to the user terminal, and complete the sample simulation of the multiple test samples on the user terminal. If it is a private cloud, the sample simulation of multiple test samples can be completed directly in the cloud.
  • determining the simulation results corresponding to the multiple test samples may specifically include: inputting the multiple test samples into the objective function respectively, and calculating and obtaining the simulation results corresponding to the multiple test samples.
  • the step of obtaining the simulation result can be performed on the cloud server or on the client, which can be specifically set according to actual usage requirements.
  • the simulation result can be sent to the other end.
  • determining the simulation results corresponding to the multiple test samples may include:
  • determining the simulation results corresponding to the multiple test samples may include:
  • the simulation results corresponding to the multiple test samples sent by the user terminal are received.
  • the method may further include: collecting the simulation results of each test sample through the SideCar (sidecar) collection method in K8S, and using etcd to store the simulation results of each test sample.
  • the technical solutions of the embodiments of the present application can be applied in many fields to solve the parameter optimization problem.
  • the allocation result of electric power resources or water conservancy resources in various regions can be used as a parameter to be processed to initiate a parameter optimization request, and the parameter to be processed can specifically be the amount of resources corresponding to each region, for example, In the power scenario, it can be the load capacity of the area.
  • detecting a parameter optimization request initiated by a target user, and determining a parameter sampling algorithm that matches the target user may include:
  • invoking the parameter sampling algorithm to generate the test sample includes:
  • the parameter sampling algorithm is invoked to generate a test sample of the parameter to be processed
  • the simulation results for outputting test samples for target users include:
  • the target resources may include power resources, water resources, data resources, etc., and the specific form and content of the resources are not too limited in the embodiments of the present application.
  • the resource element specifically represented by the parameter to be processed may be determined according to the processing target of the target resource.
  • the processing target of the target resource can be the power load capacity set in different regions, so as to minimize the total power consumption of the power grid
  • the power load capacity of different regions can be the parameter to be processed
  • the processing target can be the total power consumption of the power grid. energy calculation function.
  • the target parameter can be the power load capacity of each region under the condition of obtaining the minimum total energy consumption of the grid.
  • the processing information of the target resource can be generated, that is, according to the value of the processing parameter in the target parameter, the prompt information or setting instruction of the power load capacity of each region can be generated. Capacity settings can be made according to the power load capacity of each region.
  • the prompt information can be displayed to the user, so that the user can set the capacity of each region according to the power load capacity of each region prompted in the prompt information.
  • parameter optimization will also be involved.
  • the content or products recommended for users are also different due to differences in browsing characteristics such as users' consumption habits, areas of concern, and historical browsing behavior.
  • the user's consumption habits, areas of interest and other browsing characteristics can be parameterized to generate different browsing parameters, and the user's click target can be achieved by setting multiple browsing parameters. The characteristics are accurately analyzed, so as to find the target products with higher attention of users.
  • the solution of performing parameter sampling on a plurality of browsing parameters and performing parameter tests to determine the user's click probability can be applied to the technical solutions of the embodiments of the present application, so as to improve the test efficiency.
  • detecting a parameter optimization request initiated by a target user, and determining a parameter sampling algorithm that matches the target user includes:
  • invoking a parameter sampling algorithm to generate a test sample includes:
  • the parameter sampling algorithm is invoked to generate a test sample of the browsing parameters
  • the simulation results for outputting test samples for target users include:
  • the simulation result corresponding to the test sample of browsing parameters is output for the target user, so that the target user can configure the browsing parameters according to the test sample.
  • a flowchart of another embodiment of an information processing method provided by an embodiment of the present application may include:
  • 501 Detect a parameter optimization request initiated by the target user for the parameter sampling algorithm.
  • the user terminal can be, for example, an electronic device such as a mobile phone terminal, a computer, a notebook, a tablet computer, a virtual reality/augmented reality device, and an Internet of Things (IoT, Internet of Things) terminal.
  • an electronic device such as a mobile phone terminal, a computer, a notebook, a tablet computer, a virtual reality/augmented reality device, and an Internet of Things (IoT, Internet of Things) terminal.
  • IoT Internet of Things
  • 503 Send the sample acquisition request initiated by the target user to the server, so that the server responds to the sample acquisition request and invokes the parameter sampling algorithm to generate test samples.
  • 505 Output the simulation result of the test sample for the target user.
  • the objective function may belong to a black-box algorithm, and in this case, the parameter optimization request is a black-box optimization request.
  • the user terminal may detect the parameter optimization request initiated by the target user for the parameter sampling algorithm. Therefore, the parameter optimization request is sent to the server, and when the server receives the parameter optimization request sent by the user, it can determine the parameter sampling algorithm corresponding to the target user.
  • the client can also send a sample acquisition request initiated by the target user to the server, and when the server receives the sample acquisition request, it can call the parameter sampling algorithm to generate test samples. And based on the preset objective function, the simulation result of the test sample is determined, and the simulation result of the test sample is output for the target user.
  • the client provides the target user with the generation of sample parameters and the simulation of sample parameters. Through the interaction with the target user, the parameter optimization service can be realized, so that the target user can choose the parameter sampling algorithm for parameter optimization at any time, and improve the utilization efficiency of the black box algorithm. .
  • determining the simulation result of the test sample may include: receiving the test sample sent by the server, and inputting the test sample into the objective function for calculation to obtain the simulation result of the test sample.
  • determining the simulation result of the test sample may include: receiving the simulation result of the test sample sent by the server, wherein the simulation result is obtained by inputting the test sample into the simulation function for calculation.
  • determining the simulation result of the test sample may include:
  • test samples sent by the receiving server may include:
  • determining the simulation result of the test sample may include:
  • determining the simulation result of the test sample includes:
  • the simulation result corresponding to the test sample sent by the server is received; wherein, the simulation result is obtained by the server inputting the test sample into the preset objective function for calculation.
  • the method may further include:
  • the target sample is determined based on the test sample
  • test sample does not meet the convergence conditions, it will return to sending the sample acquisition request initiated by the target user to the server, so that the server can respond to the sample acquisition request and call the parameter sampling algorithm to generate the test sample and continue execution.
  • determining the target sample may include:
  • the simulation result may be a result index data, the higher the result index data, the better the simulation result, and the smaller the result index data, the worse the simulation result.
  • a flowchart of another embodiment of an information processing method provided by an embodiment of the present application may include:
  • 602 Detect the parameter optimization request initiated by the target user, and determine a parameter sampling algorithm matching the target user.
  • the user terminal of the target user may be, for example, electronic devices such as mobile phone terminals, computers, notebooks, tablet computers, virtual reality/augmented reality devices, Internet of Things (IoT, Internet of Things) terminals, etc.
  • electronic devices such as mobile phone terminals, computers, notebooks, tablet computers, virtual reality/augmented reality devices, Internet of Things (IoT, Internet of Things) terminals, etc.
  • IoT Internet of Things
  • the specific type of electronic device is not limited too much.
  • the client can interact with the user, and the client can communicate with a server capable of optimizing parameters.
  • the parameters to be optimized are the network depth of the machine learning model, the number of iterations, and the hyperparameters composed of the number of neurons in each layer.
  • An application example of the embodiment of the present application is introduced.
  • the user terminal is the mobile phone terminal M1 and the server is the cloud server M2 as an example.
  • the mobile phone terminal M1 can detect 701 a parameter optimization request initiated by the user for the hyperparameters composed of the network depth, the number of iterations, and the number of neurons in each layer of the machine learning model.
  • the mobile phone terminal M1 may send 702 a parameter optimization request to the cloud server M2.
  • the cloud server M2 may determine 703 a parameter sampling algorithm that matches the target user after detecting the parameter optimization request.
  • the mobile terminal M2 may detect 704 the sample acquisition request initiated by the target user, and send 705 the sample acquisition request to the cloud server M2.
  • the cloud server M2 may call the parameter sampling algorithm to generate 706 a test sample in response to the sample acquisition request initiated by the target user.
  • the cloud server may also determine 707 the simulation result of the test sample based on the preset objective function; and output 708 the simulation result of the test sample for the target user.
  • the cloud server M2 outputting the simulation result of the test sample for the target user may include: generating 7081 a prompt page for the simulation result of the test sample, and sending 7082 the prompt page to the mobile terminal M1. After receiving the prompt page, the mobile terminal M1 may display 7083 the prompt page on the display screen.
  • simulation result of the test sample may also be output in the form of a data string, short message information, or instant messaging message, etc.
  • the embodiment of the present application does not limit the specific output method of the simulation result too much.
  • the technical solutions of the embodiments of the present application can be applied to various fields such as artificial intelligence interaction, data retrieval, content recommendation, click-through rate prediction, smart factories, industrial control, etc., especially in the field of content recommendation, such as the field of e-commerce, the field of live video, Content recommendation in the social field, online education field, and resource allocation field, such as financial product allocation, power resources, water conservancy resources, supply chain allocation and other fields are more applicable.
  • the field of e-commerce In the field of e-commerce, application scenarios such as feature search, product recommendation in live broadcast scenario, content recommendation, and calculation of advertisement click-through rate are the most common.
  • the content recommendation scenario is used as an example to implement instance deployment.
  • the general recommendation process in the recommendation scenario may be as follows: parameterize the elements of the selected scenario, obtain parameters that affect the scenario, and use the parameters to identify different features of the scenario.
  • the server can display a variety of black-box algorithms for target users, and describe the functions and effects of each black-box algorithm in detail.
  • the target user can choose the parameter sampling algorithm from several black box algorithms.
  • the server obtains the parameter optimization request initiated by the target user, it can determine the parameter sampling algorithm selected by the target user.
  • the target user can initiate a sample acquisition request corresponding to the parameters matching the recommended scenario through the user terminal.
  • the server can respond to the sample acquisition request initiated by the target user and invoke the parameter sampling algorithm to generate test sample.
  • the simulation result of the test sample is determined.
  • the server determines the simulation result, it can send the simulation result to the client, and the client displays the simulation result of the test sample for the target user.
  • the target user can judge whether the test sample can be used by checking the simulation results. If the test sample is judged by the target user to meet the conditions of use, the target user can use the test sample to set the recommended elements, and recommend content to the user according to the recommended elements.
  • the system will recommend some search words (Query words) for the user.
  • Query words search words
  • the purpose of recommending search terms for users is to tap the potential purchase needs of users, increase user stickiness and increase the total number of commodity transactions.
  • the following architecture is used in the search system, combined with a deep learning Encode-Decode (encoder-decoder) network, that is, the target network, to predict the recommendation of search words.
  • the selection of the target parameter is performed on the number of search terms.
  • the parameter values of the parameters to be optimized formed by the number of search words are manually set according to human experience.
  • the parameter test of the number of search words can be automatically carried out according to the above parameter optimization process.
  • the test sample obtained by the server is the number of search words
  • the simulation result of the test sample is the number of search words after setting the number of search words.
  • the social field In the social field, it is also common to recommend content for social users and material recommendation for students. Recommendations in the social field are usually that a social user browses a social application, and the display interface of the application displays social content that the user is interested in. Usually, the recommendation in the social field is to use options such as the user's historical browsing behavior, areas of interest, and user information to form feature parameters, and a combination of different options can form parameters to be processed. When the test sample corresponding to the parameter to be processed is determined, the recommendation of social content can be performed according to the test sample. In order to find content that social users are interested in, the number and types of parameters can be optimized to obtain accurate social user content for testing.
  • the technical solutions of the embodiments of the present application may be configured in a cloud server, and a parameter optimization request may be initiated by an operation and maintenance personnel, and the operation and maintenance personnel may use user information, social type, etc. as parameters to be processed, and use a parameter sampling algorithm to continuously generate test samples.
  • the simulation results of the test parameters are obtained, so as to select the target sample from the plurality of test samples. And set the feature parameters for social content recommendation according to the target sample to achieve accurate recommendation.
  • the simulation results are output to the target user for viewing, so that when the target user selects a target sample with better simulation results from many test samples, the target sample is used to construct a machine learning model corresponding to the exponential simulation problem, and the model is trained to obtain model parameters. Then use the machine learning model obtained by training to simulate the index simulation problem with data such as the RMSE (Root Mean Squared Error) difference value of the actual stock index to simulate and calculate.
  • RMSE Root Mean Squared Error
  • power resource allocation In the field of resource allocation, take power resource allocation as an example.
  • the allocation of power resources usually involves many regions. Each region can be represented by corresponding parameters. These parameters can allocate a certain proportion of resources respectively.
  • the allocation of resources will affect the regional economy, population, environment and other information.
  • Parameters such as user type and electricity consumption can be used as parameters to be processed to initiate a parameter optimization request.
  • the parameter sampling algorithm can be used to generate a test sample of the parameter to be processed.
  • the setting of the candidate parameters can be used as the final optimization objective for the benefit/cost in the power system to determine the objective function corresponding to the optimization objective.
  • the objective function can be used to simulate the use result of the test sample to obtain the simulation result, which is Represents the value of the benefit/cost of the system. After that, by continuously generating test samples and simulating the test samples, a target sample can be selected from many test samples.
  • the power supplier can provide power resources to multiple regions at the same time, 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.
  • the target user can initiate a parameter optimization request for the respective power load capacity of each region as a parameter to be processed. After determining the parameter sampling algorithm that matches the parameter optimization request of the target user, a sample acquisition request initiated by the target user can be received to call the parameter sampling algorithm to continuously review the test samples corresponding to the test values of the power load capacity in each region.
  • the objective function can be the nonlinear constraint relationship between the power load capacity and the total energy consumption of the power grid.
  • the objective function can be used to simulate the test sample to obtain the simulation results. After that, by continuously simulating the test samples of the power load capacity of each region, the highest simulation result is selected, and the test sample corresponding to the highest simulation result can be used to generate a load capacity allocation strategy.
  • FIG. 8 is a flowchart of an embodiment of an information processing method provided by an embodiment of the present application. the method may include the following steps:
  • the allocation result of electric power resources or water conservancy resources in various regions can be used as a parameter to be processed to initiate a parameter optimization request, and the parameter to be processed can specifically be the amount of resources corresponding to each region, for example, In the power scenario, it can be the load capacity of the area.
  • the target resources may include power resources, water resources, data resources, etc., and the specific form and content of the resources are not too limited in the embodiments of the present application.
  • the resource element specifically represented by the parameter to be processed may be determined according to the processing target of the target resource.
  • the processing target of the target resource can be the power load capacity set in different regions, so as to minimize the total power consumption of the power grid
  • the power load capacity of different regions can be the parameter to be processed
  • the processing target can be the total power consumption of the power grid. energy calculation function.
  • the target parameter can be the power load capacity of each region under the condition of obtaining the minimum total energy consumption of the grid.
  • the processing information of the target resource can be generated, that is, according to the value of the processing parameter in the target parameter, the prompt information or setting instruction of the power load capacity of each region can be generated. Capacity settings can be made according to the power load capacity of each region.
  • the prompt information can be displayed to the user, so that the user can set the capacity of each region according to the power load capacity of each region prompted in the prompt information.
  • FIG. 9 is a flowchart of an embodiment of an information processing method provided by an embodiment of the present application.
  • the method may include the following steps:
  • parameter optimization will also be involved.
  • the content or products recommended for users are also different due to differences in browsing characteristics such as users' consumption habits, areas of concern, and historical browsing behavior.
  • the user's consumption habits, areas of interest and other browsing characteristics can be parameterized to generate different browsing parameters, and the user's click target can be achieved by setting multiple browsing parameters. The characteristics are accurately analyzed, so as to find the target products with higher attention of users.
  • the solution of performing parameter sampling on a plurality of browsing parameters and performing parameter tests to determine the user's click probability can be applied to the technical solutions of the embodiments of the present application, so as to improve the test efficiency.
  • the apparatus may include:
  • Request detection module 1001 used to detect a parameter optimization request initiated by a target user, and determine a parameter sampling algorithm that matches the target user.
  • the first response module 1002 is used for invoking a parameter sampling algorithm to generate a test sample in response to a sample acquisition request initiated by a target user.
  • the first simulation module 1003 is used for determining the simulation result of the test sample based on the preset objective function.
  • the first output module 1004 is used for outputting the simulation result of the test sample for the target user.
  • the first simulation module may include:
  • a first sending unit configured to send the test sample to the user terminal of the target user, so that the user terminal can input the test sample into a preset objective function, and calculate and obtain the simulation result corresponding to the test sample;
  • the first receiving unit is configured to receive the simulation result corresponding to the test sample sent by the user terminal.
  • the first simulation module may include:
  • the first calculation unit is configured to input the test sample into a preset objective function, and obtain the simulation result of the test sample by calculation.
  • the apparatus may also include:
  • a node determination unit used to determine the target node for starting the parameter sampling algorithm
  • the first response module may include:
  • a first response unit configured to send the sample acquisition request to the target node that starts the parameter sampling algorithm in response to the sample acquisition request
  • the sample acquisition unit is used to acquire the test samples generated by the parameter sampling algorithm through the target node.
  • the first computing unit may include:
  • the module establishment unit is used to establish the simulation module for the objective function
  • the module computing unit is used to input the test sample into the simulation module, and obtain the simulation result of the test sample by calculation.
  • the apparatus may further include:
  • the identification receiving module is used to receive the simulation test identification sent by the target user
  • the identification mark module is used to store the simulation result of the test sample for the target user based on the simulation test identification.
  • the request detection module may include:
  • a request receiving unit configured to receive a parameter optimization request sent by a target user; wherein the parameter optimization request includes request service information for the parameter sampling algorithm;
  • the first storage unit is configured to determine a parameter sampling algorithm for a target user to request a service in the requested service information when the data storage component has a storage operation of requesting service information.
  • the device may further include:
  • a first verification unit configured to verify the parameter optimization request of the target user based on the verification information in the requested service information, and obtain a verification result
  • the second storage unit is configured to store the requested service information in the data storage component if the verification result is that the verification is successful.
  • the first storage unit can be specifically used for:
  • the parameter sampling algorithm for the target user to request the service in the request service information is determined.
  • the apparatus may further include:
  • the convergence judgment module is used to obtain the judgment result of whether the test sample satisfies the preset convergence condition
  • a first processing module configured to determine a target sample based on the test sample if the judgment result is that the test sample satisfies the convergence condition
  • the second processing module is configured to jump to the first response module to continue execution if the judgment result is that the test sample does not meet the convergence condition.
  • the first response module may include:
  • the second response unit is configured to call the parameter sampling algorithm in batch mode to generate multiple test samples in response to the sample acquisition request initiated by the target user;
  • the first simulation module may include:
  • a first simulation unit configured to determine simulation results corresponding to a plurality of test samples based on a preset objective function
  • the first output module may include:
  • the first output unit is used for outputting simulation results corresponding to the multiple test samples respectively for the target user.
  • the first simulation module may include:
  • the simulation establishment unit is used to establish multiple simulation modules for the objective function
  • the simulation matching module from the multiple simulation modules, respectively matches the target simulation modules for multiple test samples
  • the second simulation module is used for inputting any test sample into the target simulation module corresponding to the test sample, and obtains the simulation result of the test sample by calculation, so as to obtain the simulation results corresponding to the plurality of test samples respectively.
  • the first simulation unit can be specifically used for:
  • the request detection module may include:
  • the second receiving unit is configured to receive a parameter optimization request initiated by the target user for the to-be-processed parameters of the target resource.
  • a first determining unit configured to determine a parameter sampling algorithm that matches the processing target of the target resource
  • the first response module may include:
  • a request-response unit is used to call a parameter sampling algorithm to generate a test sample of the to-be-processed parameter in response to a sample acquisition request initiated by the target user for the to-be-processed parameter;
  • the first output module may include:
  • the second output unit is configured to output the simulation result corresponding to the test sample of the parameter to be processed for the target user, so that the target user can process the target resource according to the test sample.
  • the request detection module may include:
  • a third receiving unit configured to receive a parameter optimization request for browsing parameters generated by a browsing operation initiated by a target user
  • a second determining unit configured to determine a parameter sampling algorithm that matches the browsing target of the target user
  • the first response module may include:
  • the second unit of request response is used to call the parameter sampling algorithm to generate the test sample of the browsing parameter in response to the sample acquisition request initiated by the target user for the browsing parameter;
  • the first output module may include:
  • the third output unit is configured to output the simulation result corresponding to the test sample of the browsing parameter for the target user, so that the target user can configure the browsing parameter according to the test sample.
  • the objective function belongs to a black-box algorithm; the request detection module can be specifically used for:
  • a black box optimization request initiated by the target user is detected, and a parameter sampling algorithm matching the target user is determined.
  • the information processing apparatus shown in FIG. 10 can execute the information processing method of the embodiment shown in FIG. 1 , and the implementation principle and technical effect thereof will not be described again.
  • the specific manner of each step performed by the processing component in the foregoing embodiment has been described in detail in the embodiment of the method, and will not be described in detail here.
  • the information processing apparatus shown in FIG. 10 may be configured as a server.
  • the server may include: a storage component 1101 and a processing component 1102 ; the storage component is used to store one or more computer instructions; one or more A computer instruction is invoked by a processing component;
  • Processing component 1102 can be used to:
  • Detect the parameter optimization request initiated by the target user and determine the parameter sampling algorithm that matches the target user; in response to the sample acquisition request initiated by the target user, call the parameter sampling algorithm to generate the test sample; Based on the preset objective function, determine the simulation result of the test sample ; Output the simulation result of the test sample for the target user.
  • the processing component may also execute the information processing method shown in any one of the foregoing embodiments, which is not repeated here for the sake of brevity of description.
  • the server in this embodiment of the present application may be the server of the foregoing embodiments, and may perform data or information interaction with the client corresponding to the user equipment.
  • the processing component 1102 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-mentioned methods.
  • the processing components 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) , a controller, a microcontroller, a microprocessor or other electronic components to implement the above information processing method.
  • ASICs application specific integrated circuits
  • DSPs digital signal processors
  • DSPDs digital signal processing devices
  • PLDs programmable logic devices
  • FPGAs field programmable gate arrays
  • the storage component 1101 is configured to store various types of data to support operations at the terminal.
  • the storage components can be implemented by any type of volatile or non-volatile storage devices or combinations thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EEPROM), Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic or Optical Disk.
  • SRAM Static Random Access Memory
  • EEPROM Electrically Erasable Programmable Read Only Memory
  • EEPROM Erasable Programmable Read Only Memory
  • EPROM Programmable Read Only Memory
  • PROM Programmable Read Only Memory
  • ROM Read Only Memory
  • Magnetic Memory Flash Memory
  • Flash Memory Magnetic or Optical Disk
  • the computing device may, of course, also include other components, such as input/output interfaces, communication components, and the like.
  • the input/output interface provides an interface between the processing component and a peripheral interface module, and the above-mentioned peripheral interface module may be an output device, an input device, or the like.
  • the communication components are configured to facilitate wired or wireless communication, and the like, between the computing device and other devices.
  • an embodiment of the present application further provides a computer-readable storage medium, where the storage medium can store one or more computer instructions, and when the one or more computer instructions are executed, is used to implement any one of FIG. 1 to FIG. 5 in the embodiment of the present application an information processing method.
  • FIG. 12 a schematic structural diagram of an embodiment of an information processing apparatus provided by an embodiment of the present application, the apparatus may include:
  • Request initiating module 1201 used to detect a parameter optimization request initiated by a target user for a parameter sampling algorithm.
  • the first sending module 1202 used to send the parameter optimization request provided by the target user to the server, so that the server can determine the parameter sampling algorithm that matches the target user;
  • the second sending module 1203 is configured to send the sample acquisition request initiated by the target user to the server, so that the server responds to the sample acquisition request and invokes a parameter sampling algorithm to generate test samples.
  • the second simulation module 1204 is used for determining the simulation result of the test sample based on the preset objective function.
  • the second output module 1205 is used for outputting the simulation result of the test sample for the target user.
  • the second simulation module may include:
  • the sample receiving unit is used to receive the test sample sent by the server;
  • the sample simulation unit is used to input the test sample into the objective function, and calculate and obtain the calculation result corresponding to the test sample.
  • the sample receiving unit may be specifically configured to: receive multiple test samples sent by the server; wherein, the multiple test samples are generated by the server calling the parameter sampling algorithm in batch mode.
  • the sample simulation unit can be specifically used for: inputting a plurality of test samples into the objective function respectively, and calculating and obtaining calculation results corresponding to the plurality of test samples respectively.
  • the second simulation module may include:
  • the result receiving unit is configured to receive the simulation result corresponding to the test sample sent by the server; wherein, the simulation result is obtained by the server inputting the test sample into the preset objective function for calculation.
  • the apparatus may also include:
  • the convergence judgment module is used to obtain the judgment result of whether the test sample satisfies the preset convergence condition
  • a first processing module configured to determine a target sample based on the test sample if the judgment result is that the test sample satisfies the convergence condition
  • the second processing module is configured to jump to the second sending module to continue execution if the judgment result is that the test sample does not meet the convergence condition.
  • the information processing apparatus shown in FIG. 12 can execute the information processing method of the embodiment shown in FIG. 7 , and the implementation principle and technical effects thereof will not be described again.
  • the specific manner of each step performed by the processing component in the foregoing embodiment has been described in detail in the embodiment of the method, and will not be described in detail here.
  • the information processing apparatus shown in FIG. 12 may be configured as a user equipment.
  • the server may include: a storage component 1301 and a processing component 1302; the storage component 1301 is used to store one or more computer instructions; a One or more computer instructions are invoked by processing component 1302 .
  • Processing component 1302 can be used to:
  • Detect the parameter optimization request initiated by the target user for the parameter sampling algorithm send the parameter optimization request provided by the target user to the server, so that the server can determine the parameter sampling algorithm that matches the target user; send the sample acquisition request initiated by the target user to the service terminal, so that the server responds to the sample acquisition request and invokes the parameter sampling algorithm to generate the test sample; based on the preset objective function, the simulation result of the test sample is determined; and the simulation result of the test sample is output for the target user.
  • the user equipment shown in the embodiments of the present application may specifically be used as the client terminals in the foregoing embodiments.
  • For the specific processing methods and beneficial effects of the various steps in the embodiments of the present application reference may be made to the relevant descriptions of the client terminals in the foregoing embodiments. Repeat.
  • processing component may also execute the information processing method shown in any one of the foregoing embodiments, which is not repeated here for the sake of brevity of description.
  • the processing component 1302 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-mentioned methods.
  • the processing components 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) , a controller, a microcontroller, a microprocessor or other electronic components to implement the above information processing method.
  • ASICs application specific integrated circuits
  • DSPs digital signal processors
  • DSPDs digital signal processing devices
  • PLDs programmable logic devices
  • FPGAs field programmable gate arrays
  • the storage component 1301 is configured to store various types of data to support operations at the terminal.
  • the storage components can be implemented by any type of volatile or non-volatile storage devices or combinations thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EEPROM), Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Magnetic or Optical Disk.
  • SRAM Static Random Access Memory
  • EEPROM Electrically Erasable Programmable Read Only Memory
  • EEPROM Erasable Programmable Read Only Memory
  • EPROM Programmable Read Only Memory
  • PROM Programmable Read Only Memory
  • ROM Read Only Memory
  • Magnetic Memory Flash Memory
  • Flash Memory Magnetic or Optical Disk
  • the computing device may, of course, also include other components, such as input/output interfaces, communication components, and the like.
  • the input/output interface provides an interface between the processing component and a peripheral interface module, and the above-mentioned peripheral interface module may be an output device, an input device, or the like.
  • the communication components are configured to facilitate wired or wireless communication, and the like, between the computing device and other devices.
  • an embodiment of the present application also provides a computer-readable storage medium, where the storage medium can store one or more computer instructions, and when the one or more computer instructions are executed, is used to realize the information processing shown in FIG. 6 in the embodiment of the present application method.
  • the device embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place , or distributed to multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Mathematical Physics (AREA)
  • Mathematical Optimization (AREA)
  • Data Mining & Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Software Systems (AREA)
  • Marketing (AREA)
  • Human Resources & Organizations (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Algebra (AREA)
  • Operations Research (AREA)
  • Evolutionary Computation (AREA)
  • Game Theory and Decision Science (AREA)
  • Databases & Information Systems (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Quality & Reliability (AREA)
  • Geometry (AREA)
  • Tourism & Hospitality (AREA)
  • Computer Hardware Design (AREA)
  • Educational Administration (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Debugging And Monitoring (AREA)
  • Stored Programmes (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请实施例提供一种信息处理方法及装置、服务器及用户设备，该方法包括：检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；基于预设目标函数，确定所述测试样本的仿真结果；为所述目标用户输出所述测试样本的仿真结果。本申请实施例提高了参数优化效率。

Description

信息处理方法及装置、服务器及用户设备
本申请要求2021年01月25日递交的申请号为202110099097.6、发明名称为“信息处理方法及装置、服务器及用户设备”中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及电子设备技术领域，尤其涉及一种信息处理方法及装置、服务器及用户设备。
背景技术
黑盒算法是一种为了解决较为复杂、不能使用明确的数学计算式进行建模的问题而建立的算法模型，黑盒算法的数学模型通常是未知的，将对应的输入数据输入到黑盒算法的目标函数中，计算获得该输入数据对应的输出数据。在实际计算过程中，黑盒算法具体的目标函数是未知的。由于黑盒算法的服务类型以及可选择性很高，黑盒算法可以应用于诸如电子商务、金融、电力等诸多领域。
现有技术中，开发者可以基于AutoML(Automated machine learning，自动机器学习)框架，实现黑盒优化。黑盒优化的主要过程是，针对待优化参数，通过参数采样算法不断产生新的样本，再采用黑盒的目标函数对样本的使用效果进行评估，获得样本的仿真结果，之后，不断利用参数采样算法产生新样本并进行仿真评估，以尽快产生接近仿真结果满足预定条件的目标样本，该目标样本以及目标样本对应的仿真结果即为此次黑盒优化的优化结果。
但是，现有的黑盒优化由开发者开发之后，主要在本地执行黑盒优化过程，导致黑盒优化效率较低。
发明内容
有鉴于此，本申请实施例提供一种信息处理方法及装置、服务器及用户设备，用以解决现有技术中由于在本地执行黑盒优化过程导致黑盒优化效率较低的技术问题。
第一方面，本申请实施例提供一种信息处理方法，包括：
检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
基于预设目标函数，确定所述测试样本的仿真结果；
为所述目标用户输出所述测试样本的仿真结果。
第二方面，本申请实施例提供一种信息处理方法，包括：
检测目标用户针对参数采样算法发起的参数优化请求；
发送目标用户提供的参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；
发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；
基于预设目标函数，确定所述测试样本的仿真结果；
为所述目标用户输出所述测试样本的仿真结果。
第三方面，本申请实施例提供一种信息处理方法，包括：
响应于调用信息处理接口的请求，确定所述信息处理接口对应的处理资源；
利用所述信息处理接口对应的处理资源执行如下步骤：
检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
基于预设目标函数，确定所述测试样本的仿真结果；
为所述目标用户输出所述测试样本的仿真结果。
第四方面，本申请实施例提供一种信息处理装置，包括：
请求检测模块，用于检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
第一响应模块，用于响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
第一仿真模块，用于基于预设目标函数，确定所述测试样本的仿真结果；
第一输出模块，用于为所述目标用户输出所述测试样本的仿真结果。
第五方面，本申请实施例提供一种信息处理装置，包括：
第一发送模块，用于发送目标用户提供的参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；
第二发送模块，用于发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；
第二仿真模块，用于基于预设目标函数，确定所述测试样本的仿真结果；
第二输出模块，用于为所述目标用户输出所述测试样本的仿真结果。
第六方面，本申请实施例提供一种服务器，包括：存储组件以及处理组件；所述存储组件用于存储一条或多条计算机指令；所述一条或多条计算机指令被所述处理组件调用；
所述处理组件用于：
检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；基于预设目标函数，确定所述测试样本的仿真结果；为所述目标用户输出所述测试样本的仿真结果。
第七方面，本申请实施例提供一种用户端，包括：存储组件以及处理组件；所述存储组件用于存储一条或多条计算机指令；所述一条或多条计算机指令被所述处理组件调用；
所述处理组件用于：
检测目标用户针对参数采样算法发起的参数优化请求；发送目标用户提供的参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；基于预设目标函数，确定所述测试样本的仿真结果；为所述目标用户输出所述测试样本的仿真结果。
本申请实施例，可以接收目标用户发起的参数优化请求，并确定与目标用户相匹配的参数采样算法。之后可以响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。测试样本可以用于在目标函数的计算方式中，确定仿真结果，从而为目标用户输出测试样本的仿真结果。通过与目标用户的交互，实现参数优化的上云服务，以实现云端的快速参数优化，提高参数优化效率。另外，通过获取用户的参数优化请求以及样本获取请求，从而获取用户的服务需求，并通过对用户的服务需求作出有效响应以及反馈，为用户有效提供黑盒服务，实现黑盒服务的扩展应用，提高黑盒算法的利用效率。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本申 请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例提供的一种信息处理方法的一个实施例的流程图；
图2为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图3为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图4为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图5为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图6为本申请实施例提供的一种信息控制方法的又一个实施例的流程图；
图7为本申请实施例提供的一种信息处理方法的一个应用示例图；
图8为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图9为本申请实施例提供的一种信息处理方法的又一个实施例的流程图；
图10为本申请实施例提供的一种信息处理装置的一个实施例的结构示意图；
图11为本申请实施例提供的一种服务器的一个实施例的结构示意图；
图12为本申请实施例提供的一种信息处理装置的又一个实施例的结构示意图；
图13为本申请实施例提供的一种用户设备的一个实施例的结构示意图。
具体实施方式
为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
在本申请实施例中使用的术语是仅仅出于描述特定实施例的目的，而非旨在限制本申请。在本申请实施例和所附权利要求书中所使用的单数形式的“一种”、“所述”和“该”也旨在包括多数形式，除非上下文清楚地表示其他含义，“多种”一般包含至少两种，但是不排除包含至少一种的情况。
应当理解，本文中使用的术语“和/或”仅仅是一种描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。另外，本文中字符“/”，一般表示前后关联对象是一种“或”的关系。
取决于语境，如在此所使用的词语“如果”、“若”可以被解释成为“在……时”或“当……时”或“响应于确定”或“响应于识别”。类似地，取决于语境，短语“如果确定”或“如果识别(陈述的条件或事件)”可以被解释成为“当确定时”或“响应于确定”或“当识别(陈述的条件或事件)时”或“响应于识别(陈述的条件或事件)”。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的商品或者系统不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种商品或者系统所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的商品或者系统中还存在另外的相同要素。
本申请实施例的技术方案可以应用于参数寻优过程中，通过采用提供对外的黑盒服务，可以实现黑盒优化的多用户应用，扩展黑盒优化的适用范围，提高黑盒优化的利用效率。
现有技术中，开发者基于AutoML框架开发黑盒算法之后，可以使用增加的黑盒优化服务。黑盒优化的主要过程是，采用黑盒算法不断产生新样本，并使用模拟用户实际计算效果的仿真函数，也即目标函数对产生的新样本进行使用效果的仿真，获得该样本的仿真结果。通过不断产生新样本，不断对新产生的样本进行效果仿真，以获得从获得的诸多样本中选择使用效果最高的样本作为全局目标解。但是现有的黑盒算法，或者黑盒优化策略多是开发者内部使用，导致黑盒优化的应用范围较窄，黑盒算法的利用率不高。
本申请实施例中，可以检测目标用户发起的参数优化请求，确定与该目标用户相匹配的参数采样算法。通过提供对外的黑盒服务，可以使得目标用户可以随时使用黑盒算法进行参数优化。在响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本之后，可以基于预设目标函数，确定该测试样本的仿真结果，自动为目标用户生成测试样本，并完成测试样本的仿真计算，仿真结果通常为测试样本的使用效果的估计结果，以对测试样本进行准确评价，从而为目标用户输出测试样本的仿真结果之后，可以供目标用户查看，并确定该测试样本是否可用，实现对测试样本的效果测试。通过与目标用户的交互，实现参数优化的上云服务，以实现云端的快速参数优化，提高参数优化效率。另外，通过为用户有效提供黑盒服务，实现黑盒服务的扩展应用，提高黑盒算法的利用效率。
下面将结合附图对本申请实施例进行详细描述。
如图1所示，为本申请实施例提供的一种信息处理方法的一个实施例的流程图，方 法可以包括以下几个步骤：
101：检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法。
本申请实施例提供的信息处理方法可以应用于服务端，服务端是相对于用户端而言的，服务端与用户端是对应于C/S(Client-Server)架构而言。在实际应用中，服务端例如可以为计算机、普通服务器、云端服务器、超级个人计算机、笔记本电脑、平板电脑等。用户端可以配置于用户设备上，用户端例如可以为手机、平板电脑、计算机、虚拟现实设备、增强现实设备、可穿戴设备等。本申请实施例对服务端以及用户端的具体类型并不作出过多限定。
可选地，服务端中可以配置有多个参数采样算法。不同参数采样算法的原理不同，但是结果相同，均可以用于产生测试样本。一般参数采样算法可以用于产生待优化参数的样本，并采用预设的目标函数来对样本的使用效果进行仿真，以预估该样本的使用效果，获得样本的仿真结果。通过不断产生样本，获得样本的仿真结果可以筛选仿真结果评价最高的目标样本，以将目标样本作为参数优化的结果输出，完成参数优化。
目标用户可以通过用户端发送参数优化请求至服务端。服务端可以接收目标用户的用户端发送的参数优化请求。参数优化请求可以由目标用户基于待优化的参数生成。以电子商务领域中的内容推荐为例，在目标用户浏览电子商务网站时，网站可以为目标用户推送目标用户感兴趣的内容，但是，由于内容的种类包括多种，不同类型的内容对点击率的贡献不同，因此，在对各个种类的推荐内容的比例可以作为待优化参数，进行参数优化。
服务端可以配置有多个候选采样算法，目标用户可以在参数优化请求的多个黑盒算法中指定参数采样算法。此时，参数优化请求中可以包括参数采样算法的算法标识信息。算法标识信息例如可以为算法名称。
此外，参数优化请求中还可以包括待优化参数的样本所对应的样本类型、取值范围、数量等参数信息。在一些可选实施例中，参数采样算法可以同时生成待优化参数的多个样本，以同时对多个样本进行评估，提高参数优化效率。为了标识不同用户的参数优化请求，参数优化请求中还可以包括为目标用户的此次参数优化试验设置的试验标识，以区别于不同参数的优化过程。目标用户还可以使用用户名以及密码等用户信息登陆服务器，以获得使用参数优化服务的权限，实现对用户的有效管理。
可选地，在实际应用中，服务端接收到参数优化请求时，可以将参数优化请求中的参数信息、用户信息、试验标识等信息进行存储，以便于对参数优化服务的过程以及结 果进行准确追踪，提高信息处理效率。
102：响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
服务端确定与目标用户相匹配的参数采样算法之后，可以向目标用户反馈启动参数优化的启动提示信息。目标用户发送参数优化请求之后，可以接收服务端反馈的参数优化对应的启动提示消息。当目标用户确认黑盒算法启动时，可以发送样本获取请求至服务端。服务端可以接收目标用户发送的样本获取请求，响应该样本获取请求，调用参数采样算法生成测试样本。
测试样本可以是为待优化参数设置的参数值，具体可以为参数采样算法为待优化参数生成的一个候选解，该候选解在满足优化条件时，构成待优化参数的最终优化结果。
103：基于预设目标函数，确定测试样本的仿真结果。
目标函数可以对测试样本的使用效果或者使用结果进行仿真，获得仿真结果。目标函数可以通过待优化参数的优化目标确定，例如，在内容推荐场景中，待优化参数可以为与推荐内容对应的参数，测试样本即为根据待优化参数查找到的向浏览用户展示的推荐内容，用户点击推荐内容的概率越高，说明此次内容推荐越有效，因此，可以将点击率的预测模型作为目标函数，预测浏览用户对该推荐内容的点击率，预测获得的点击率即为该测试样本的仿真结果。
104：为目标用户输出测试样本的仿真结果。
可选地，为目标用户输出测试样本的仿真结果可以包括：发送测试样本的仿真结果至目标用户的用户端，以供用户端显示测试样本的输出结果。
作为一种可能的实现方式，服务端将测试样本的仿真结果发送至目标用户的用户端时，可以以网页、即时通讯消息、短信息等消息形式发送至目标用户的用户端。
本申请实施例中，可以检测目标用户发起的参数优化请求，确定与该目标用户相匹配的参数采样算法。通过提供对外的黑盒服务，可以使得目标用户可以随时使用黑盒算法进行参数优化。在响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本之后，可以基于预设目标函数，确定该测试样本的仿真结果，自动为目标用户生成测试样本，并完成测试样本的仿真计算。通过与目标用户的交互，实现参数优化的上云服务，以实现云端的快速参数优化，提高参数优化效率。另外，通过可以获取用户的黑盒服务请求参数优化请求以及样本获取请求，以从而获取用户的服务需求，并通过对用户的服务需求作出有效响应以及反馈，为用户有效提供黑盒服务，实现黑盒服务的扩展应用，提高黑盒算法的利用效率。
在某些实施例中，测试样本的仿真结果通常为基于目标函数对测试样本的使用效果进行评估获得的评估结果，通过仿真结果可以对测试样本的使用效果进行准确评价，从而为目标用户输出测试样本的仿真结果之后，目标用户查看测试样本的仿真结果，以直接通过仿真结果确定该测试样本是否满足使用要求，实现对测试样本的使用效果的准确提示。例如，在前述的内容推荐示例中，以点击率计算模型作为目标函数时，计算获得的点击率越高，说明推荐内容越有效，当点击率达到目标用户设置的阈值时，目标用户即可以确认测试样本满足使用条件，此时，可以将该测试样本提供给推荐系统，以使推荐系统基于测试样本生成推荐内容向使用该推荐系统的用户推荐。
本申请实施例中，可以面向系统开发者、学者、学生、系统设计者、企业用户等诸多类型的目标用户开放，以向不同类型的目标用户提供参数优化服务。为目标用户有效提供参数优化服务时，目标用户可以不断发起样本获取请求，服务端可以持续响应样本获取请求，为目标用户提供测试样本，使得目标用户获知各个测试样本的仿真结果，从中选择满足其使用条件的目标样本。向不同目标用户提供参数优化服务，实现参数优化服务的扩展应用，令更多用户使用参数优化服务，提高了参数优化的利用效率。
本申请实施例中，目标函数具体可以属于黑盒算法，不能使用明确的数学公式表示，目标函数的具体的数学模型通常是未知的，此时，参数优化请求即可以包括黑盒优化请求。
作为一个实施例，检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法可以包括：
检测目标用户发起的黑盒优化请求，确定与目标用户相匹配的参数采样算法。
本申请实施例的技术方案可以应用于云服务器中，云服务器可以包括公有云以及私有云。其中，共有云可以为第三方服务上云用户提供的能够使用的云，一般通过网络使用，实现共享资源服务，可以为开放的公有网络提供服务。私有云是为一个客户单独使用而构建的，因而提供对数据、安全性和服务质量的最有效控制。因此，私有云的安全性要高于公有云。为了数据的安全性起见，用户使用云端配置的信息处理方法时，公有云以及私有云的处理流程可以有所不同。
采用目标函数对测试样本进行仿真时，涉及到用户信息或者数据较多。对于私有云而言，采用目标函数对测试样本进行仿真时，由于私有云为用户单独服务，私密性较高，因此，可以将测试样本的仿真过程在私有云完成，以减少用户端的处理压力。对于公有云而言，采用目标函数对测试样本进行仿真时，由于公有云为很多用户提供服务，不具 备私密性，因此，为了确保数据的安全性，可以将测试样本的仿真过程在用户端完成。
在一种可能的设计中，共有云以及私有云可以是基于K8S的底层服务实现。K8S(Kubernets，容器集群管理系统)可以用于自动部署、扩展和管理容器化应用程序。集群是一组节点，云服务器，无论是公有云还是私有云可以看着是一个服务集群，服务集群中可以由多个节点构成，这些节点上安装了K8S平台，以便于使用K8S对集群中的节点进行管理。本申请实施例中的信息处理方法可以适用于K8S平台，具体可以配置于K8S平台中创建的Custom resource controller(CRD，自定义资源控制器)上，也即，K8S中的CRD可以作为处理组件实现信息处理方法。
本申请实施例的技术方案，可以在K8S底层服务的基础上进行配置，可以将各个步骤的功能模块化，然后配置于节点上，并在功能模块的基础上提供对外服务的接口，通过对外服务接口可以实现与用户端的数据或者信息交互，为用户提供参数优化服务。例如，云服务器中可以在K8S的基础上，配置网关、数据存储组件、黑盒算法组件、早期停止组件、web(World Wide Web，全球广域网)页面、以及样本收集系统、数据库等组件，通过这些组件的配合完成本申请实施例的技术方案。
在一些实施例中，网关可以为K8S的Rest(Representational State Transfer，表症征状态转移)API(Application Programming Interface，应用程序编程接口)网关API Server(应用程序编程接口服务)；数据存储组件可以为K8S的分布式存储组件Etcd(存储集群状态信息)；黑盒算法组件可以为开发人员提供的核心算法组件Suggestion Service(建议服务)，例如可以包括：random search(随机搜索)/grid search(超参寻优)/cmaes(极大似然估计)/pma(Partitioning Around Medoid,围绕中心点的划分)中的一种或多种；早期停止(Early Stop)组件可以为通过早期停止技术对过程不理想的参数仿真，及早停止，减少资源消耗，加速试验进度；web页面主要用于对外向用户输出测试样本以及测试样本的仿真结果，便于用户及时查看；样本收集系统(Metric System)可以为收集pod(调度单元)内的黑盒测试样本；数据库可以为MySQL等数据库，用于存储测试样本以及仿真过程产生的数据以及仿真结果等。
在云服务器为公有云时，测试样本的仿真过程可以在用户端完成，以减少数据泄露可能，提高数据安全性。如图2所示，为本申请实施例提供的一种信息处理方法的又一个实施例的流程图，该方法可以包括以下几个步骤：
201：检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法。
本申请实施例中的部分步骤与图1所示的部分步骤相同，为了描述的简洁性，在此 不再赘述。本申请实施例中，参数优化请求可以由用户端发送至服务端，服务端可以提供对外通信的接口，也即信息处理接口，通过信息处理接口接收用户端发送的参数优化请求。
可选地，用户端向服务端发起参数优化请求时，可以通过在用户端输入的算法信息、参数信息、用户信息和/或校验信息等信息构成的请求服务信息可以以SDK(Software Development Kit，软件开发工具包)的形式发送至服务端，服务端的信息处理接口可以接收以SDK形式封装的参数优化请求，获得请求服务信息对应的参数采样算法。
202：响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
203：发送测试样本至目标用户的用户端，以供用户端将测试样本输入预设的目标函数，计算获得测试样本对应的仿真结果。
将测试样本发送至目标用户的用户端，由用户端完成测试样本的仿真工作，可以减少数据泄露风险，提高数据安全性。
可选地，发送测试样本至目标用户的用户端之前，该方法还可以包括：接收目标用户的用户端发送的样本测试请求，响应该样本测试请求，将测试样本发送至目标用户的用户端。
在又一种可能的设计中，可以对目标用户进行测试提示，例如，可以生成测试提示信息，该测试提示信息可以包括对在公有云进行样本测试的风险提示信息，例如以文字信息的方式提醒目标用户“是否在公有云上执行测试样本的仿真计算，存在数据泄露的风险”。此时，用户端接收到风险提示信息之后，可以为目标用户输出风险提示信息，以使得目标用户查看风险提示信息，并对风险提示信息作出相应的反馈操作。在一些实施例中，目标用户可以直接在用户端进行测试样本的仿真作出确认操作，服务端检测到该确认操作时，可以发送测试样本至目标用户的用户端。
204：接收用户端发送的测试样本对应的仿真结果。
205：为目标用户输出测试样本的仿真结果。
可选地，为目标用户输出测试样本的仿真结果可以包括：控制目标用户输出测试样本的仿真结果，例如，服务端接收到测试样本的仿真结果时，可以对仿真结果进行验证，验证成功之后，可以输出仿真结果的输出控制信息至用户端，用户端即可以响应仿真结果的输出控制信息，输出仿真结果。
本申请实施例中，检测目标用户发起的参数优化请求，可以确定与目标用户相匹配的参数采样算法，可以响应于目标用户发起的样本获取请求，调用参数采样算法生成测 试样本。从而发送测试样本至目标用户的用户端，以供用户端将测试样本输入预设的目标函数，计算获得测试样本对应的仿真结果。将测试样本的测试工作放置于用户端，减少数据传输，提高交互效率。接收到用户端发送的测试样本对应的仿真结果之后，可以为目标用户输出测试样本的仿真结果。
在云服务器为私有云时，测试样本的仿真过程可以在用户端完成，以确保数据的安全性。如图3所示，为本申请实施例提供的一种信息处理方法的又一个实施例的流程图，该方法可以包括以下几个步骤：
301：检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法。
本申请实施例中的部分步骤与图1、图2所示实施例的部分步骤相同，为了描述的简洁性，在此不再赘述。本申请实施例中，参数优化请求可以由用户端发送至服务端，服务端可以提供对外通信的接口，也即信息处理接口，通过信息处理接口接收用户端发送的参数优化请求。
可选地，用户端向服务端发起参数优化请求时，可以通过在用户端输入的算法信息、参数信息、用户信息和/或校验信息等信息构成的请求服务信息可以以声明式API(Application Programming Interface，应用程序接口)的方式发送至服务端。声明式指的是用户端提交一个对请求服务信息定义好的API对象来声明所期望的请求服务信息的具体内容以及类型。声明式API可以为K8S服务的项目编排能力作出贡献，使得参数优化请求可以按照一定的对象格式在用户端与服务端之间交互，实现信息的准确传输。
服务端接收到参数优化请求时，可以按照参数优化请求中定义的信息类型以及信息数量进行解析，获得参数优化请求中的相关信息，从而确定与目标用户相匹配的参数采样算法。
由前述实施例中的描述可知，黑盒算法组件中实际可以包括多个经典算法，参数采样算法为其中一种，在一些实施例中，参数采样算法可以直接由目标用户指定，也即，此时，参数优化请求中可以包括参数采样算法的算法标识信息。在又一些实施例中，参数采样算法可以由服务端按照目标用户提供的参数信息确定，以为该参数信息选择更匹配的黑盒算法，提高参数采样算法与当前的参数优化需求的适用性。
302：响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
303：将测试样本输入预设的目标函数，计算获得测试样本的仿真结果。
可选地，将测试样本输入预设的目标函数，计算获得测试样本的仿真结果可以包括：确定目标函数的调度执行单元(Pod)，将测试样本输入到该调度执行单元中的目标函数， 由该调度执行单元计算获得测试样本的仿真结果。调度执行单元为K8S服务提供的计算模块。
304：为目标用户输出测试样本的仿真结果。
本申请实施例中，服务端检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法时，可以响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。而将测试样本输入预设的目标函数，计算获得测试样本的仿真结果，为目标用户输出测试样本的仿真结果。将测试样本的仿真过程在服务端完成，可以减少用户端的处理压力，另外，由于服务端的计算效率一般高于用户端，可以提高测试样本的处理效率。通过提供对外的黑盒服务，可以使得目标用户可以随时使用黑盒算法进行参数优化，提高黑盒算法的利用效率。
可选地，服务端为云服务器时，云服务器可以由多个节点集成获得，参数采样算法可以由其中一个节点运行。作为一个实施例，检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法之后，还包括：
确定启动参数采样算法的目标节点；
响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本包括：
响应于样本获取请求，将样本获取请求发送至启动参数采样算法的目标节点；
通过目标节点获取参数采样算法生成的测试样本。
可选地，云服务器是基于规模化的服务器集群，可以由多个计算节点构成。通过节点查询可以确定启动参数采样算法中的目标节点。可选地，确定启动参数采样算法的目标节点可以包括：通过反向代理从多个节点中查询启动黑盒算法的目标节点。作为一种可能的实现方式，可以检测各个节点中的黑盒算法的运行状态。不同节点中可以分别配置黑盒算法，任一个节点中可以配置一个或多个黑盒算法。
在以K8S为基础服务的云服务器中，参数采样算法可以在云服务器中的某个节点中运行，用户端实际可以从运行参数采样算法的目标节点上获取测试样本。通常，可以通过路由器实现用户端与目标节点之间的通信。
作为一种可能的实现方式，将测试样本输入预设的目标函数，计算获得测试样本的仿真结果可以包括：
为目标函数建立仿真模块；
将测试样本输入仿真模块，计算获得测试样本的仿真结果。
目标函数在运行过程中需要依赖于计算模块的支持，因此，为了令目标函数能够正 常对测试样本进行仿真计算，为目标函数建立仿真模块。在K8S服务中，仿真模块可以为Pod。
服务端可以同时接收多个用户端分别发送的参数优化请求，从而同时服务多个用户端，或者一个用户端针对不同的优化需求，建立不同的参数优化请求，服务端还可以同时接收一个用户端发送的多个参数优化请求，因此，为了区分不同的黑盒优化过程，目标用户可以在发起参数优化请求时，同时向服务端发送仿真试验标识，以区别不同的黑盒服务。
因此，作为一种可能的实现方式，该方法还可以包括：
接收目标用户发送的仿真试验标识；
基于仿真试验标识，为目标用户存储测试样本的仿真结果。
可选地，接收目标用户发送的仿真试验标识具体可以是检测目标用户发起的参数优化请求时，接收目标用户发送的仿真试验标识。仿真试验标识用于标识目标用户发起的参数优化请求中的测试样本的仿真试验。例如，当目标用户A与目标用户B同时发起参数优化请求时，目标用户A发起参数优化请求时，还可以发送仿真试验标识a至云服务器，目标用户B发起参数优化请求时，还可以发送仿真试验标识b至云服务器。当目标用户A发起样本获取请求时，可以通过仿真试验标识a标识为目标用户A生成的测试样本以及测试样本的仿真结果。当目标用户B发起样本获取请求时，可以通过仿真试验标识b标识为目标用户B生成的测试样本以及测试样本的仿真结果，以区别于为目标用户A生成的使用仿真试验标识a标识的测试样本以及仿真结果。
在实际应用中，服务端提供黑盒优化服务时，可以基于用户指定的参数采样算法启动黑盒优化过程。如图4所示，为本申请实施例提供的一种信息处理方法的又一个实施例的流程图，该方法可以包括：
401：接收目标用户发起的参数优化请求。
其中，参数优化请求中包括针对参数采样算法的请求服务信息。
可选地，接收目标用户发起的参数优化请求之前，该方法还可以包括：发送多个黑盒算法至目标用户的用户端，以供该用户端为目标用户输出多个黑盒算法，并从多个黑盒算法中选择参数采样算法。请求服务信息可以包括算法名称、算法标识等可以用于识别黑盒算法的信息。
402：检测数据存储组件存在请求服务信息的存储操作时，确定目标用户在请求服务信息中请求服务的参数采样算法。
用户端输入的算法信息、参数信息、用户信息和/或校验信息等信息构成的请求服务信息。服务端对参数优化请求执行验证等操作之后，可以将请求服务信息进行存储。在K8S服务中，通常是采用etcd(分布式一致性协议)存储请求服务信息。Etcd是kubenetes中开发的分布式存储组件，可以存储服务器集群中网络配置和对象的状态信息。
可选地，etcd将请求服务信息的存储操作具体可以是检测etcd的存储动作，当etcd存在存储动作时，可以向处理组件发起存储消息，以使得存储组件及时确定数据存储组件存在请求服务信息的存储操作。
403：响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
404：基于预设目标函数，确定测试样本的仿真结果。
405：为目标用户输出测试样本的仿真结果。
本申请实施例中部分步骤与图1～图3所示实施例中部分步骤相同，为了描述的简洁性考虑，在此不再赘述。
本申请实施例中，目标用户发起的参数优化请求中可以包括目标用户指定的参数采样算法的请求服务信息。服务端可以检测数据存储组件的存储行为，当检测到数据存储组件存在请求服务信息的存储操作时，可以确定目标用户在请求服务信息中请求服务的参数采样算法。通过用户指定黑盒服务算法，实现针对性服务，增强用户对黑盒算法的可控性。确定参数采样算法之后，可以响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。基于预设目标函数，确定测试样本的仿真结果，从而为目标用户输出测试样本的仿真结果。通过提供对外的黑盒服务，可以使得目标用户可以随时选择参数采样算法进行参数优化，提高黑盒算法的利用效率。
在实际应用中，请求服务信息中还可以包括校验信息。该校验信息可以包括目标用户的身份校验信息、权限校验信息等。
在一种可能的设计中，接收目标用户发送的参数优化请求之后，该方法还可以包括：
基于请求服务信息中的校验信息，对目标用户的参数优化请求进行校验，获得校验结果；
若校验结果为校验成功，则将请求服务信息存储至数据存储组件。
可选地，校验信息可以包括目标用户的身份校验信息以及权限校验信息，基于请求服务信息中的校验信息，对目标用户的参数优化请求进行校验，获得校验结果可以包括基于请求服务信息中的身份校验信息对目标用户的用户身份进行校验，获得身份校验结果；
基于请求服务信息中的权限校验信息对目标用户的使用权限进行校验，获得权限校验结果。
综合身份校验结果以及权限校验结果，确定校验结果。
进一步，可选地，综合身份校验结果以及权限校验结果，确定校验结果可以包括：若身份校验结果为校验成功以及权限校验结果为校验成功，则确定校验结果为校验成功。若身份校验结果为校验失败或者权限校验结果为校验失败，则确定校验结果为校验失败。
在某些实施例中，若校验结果为校验失败，则反馈校验失败的提示信息至目标用户的用户端，以供用户端为目标用户输出校验失败的提示信息，提醒用户进行身份信息或者权限信息的失败原因，并根据失败原因进行信息修改或者权限修改，重新进行服务申请。
作为又一个实施例，检测数据存储组件存在请求服务信息的存储操作时，确定目标用户在请求服务信息中请求服务的参数采样算法可以包括：
通过资源控制器获取数据存储组件在存储请求服务信息时，发起的存储通知消息；
基于存储通知消息，确定目标用户在请求服务信息中请求服务的参数采样算法。
服务端中可以配置资源控制器，并通过资源控制器获取数据存储组件在存储请求服务信息时，发起的存储通知消息。通过存储通知消息，可以确定目标用户在请求服务信息中请求服务的参数采样算法。
作为又一个实施例，响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本之前，该方法还可以包括：
接收目标用户针对参数采样算法发起的样本获取请求。
在某些实施例中，基于预设目标函数，确定测试样本的仿真结果之后，该方法还可以包括：
获取测试样本是否满足预设收敛条件的判断结果；
如果判断结果为测试样本满足收敛条件，则基于测试样本，确定目标样本。
如果判断结果为测试样本不满足收敛条件，则返回至响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本继续执行。
可选地，基于测试样本，确定目标样本包括：
获取该测试样本之前的N次样本获取请求分别对应的历史测试样本，并基于该测试样本以及N个历史测试样本分别对应的仿真结果，从该测试样本以及N个历史测试样本中选择仿真结果最高的目标样本。
仿真结果具体可以为一结果指标数据，结果指标数据越高，仿真结果越好，仿真结果越低，仿真结果越差。
在某些实施例中，测试样本是否满足预设收敛条件可以通过测试样本所对应的迭代次数是否达到预设的迭代次数阈值，如果是，则确定测试样本达到预设收敛条件，如果否，则确定测试样本未达到预设收敛条件。
在某些实施例中，测试样本是否满足预设收敛条件可以通过测试样本的仿真结果是否满足预设的结果阈值，如果是，则确定测试样本达到预设收敛条件，如果否，则确定测试样本未达到预设收敛条件。仿真结果是否满足预设结果阈值具体可以包括仿真结果大于预设结果阈值，或者，仿真结果小于预设结果阈值。仿真结果大于结果阈值还是小于结果阈值具体可以根据目标函数的计算含义确定。
为了提高参数处理效率，可以以批处理方式一次产生多个测试样本。
在某些实施例中，响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本可以包括：
响应于目标用户发起的样本获取请求，以批处理方式调用参数采样算法生成多个测试样本。
基于预设目标函数，确定测试样本的仿真结果可以包括：
基于预设目标函数，确定多个测试样本分别对应的仿真结果。
为目标用户输出测试样本的仿真结果可以包括：
为目标用户输出多个测试样本分别对应的仿真结果。
可选地，基于预设目标函数确定多个测试样本分别对应的仿真结果时，如果是，公有云，可以将多个测试样本发送至用户端，在用户端完成多个测试样本的样本仿真。如果是私有云，可以直接在云端完成多个测试样本的样本仿真。
可选地，基于预设目标函数，确定多个测试样本分别对应的仿真结果具体可以包括：将多个测试样本分别输入目标函数，计算获得多个测试样本分别对应的仿真结果。
在某些实施例中，仿真结果的获取步骤可以在云服务器执行，也可以在用户端执行，具体可以根据实际使用需求设定。服务端或者用户端中的任一端获得仿真结果时，可以将仿真结果发送至另外一端。
在一种可能的设计中，基于预设目标函数，确定多个测试样本分别对应的仿真结果可以包括：
为目标函数建立多个仿真模块。
从多个仿真模块中，分别为多个测试样本匹配目标仿真模块；
将任一个测试样本输入该测试样本对应的目标仿真模块，计算获得测试样本的仿真结果，以获得多个测试样本分别对应的仿真结果。
在又一种可能的设计中，基于预设目标函数，确定多个测试样本分别对应的仿真结果可以包括：
发送多个测试样本至目标用户的用户端，以供用户端将多个测试样本分别输入预设的目标函数，计算获得多个测试样本分别对应的仿真结果；
接收用户端发送的多个测试样本分别对应的仿真结果。
在某些实施例中，该方法还可以包括：通过K8S中的SideCar(挎斗)采集方式，采集各个测试样本的仿真结果，并利用etcd存储各个测试样本的仿真结果。
本申请实施例的技术方案可以应用诸多领域中，以解决参数优化问题。在电力资源、水利资源的分配过程中，可以将电力资源或者水利资源在各个地区的分配结果作为一待处理参数发起参数优化请求，待处理参数具体可以为各个区域所对应的资源量，例如，电力场景中可以为区域的负载容量。
作为一个实施例，检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法可以包括：
接收目标用户针对目标资源的待处理参数发起的参数优化请求；
确定与目标资源的处理目标相匹配的参数采样算法；
响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本包括：
响应于目标用户针对待处理参数发起的样本获取请求，调用参数采样算法生成待处理参数的测试样本；
为目标用户输出测试样本的仿真结果包括：
为目标用户输出待处理参数的测试样本对应仿真结果，以供目标用户按照测试样本处理目标资源。
可选地，目标资源可以包括电力资源、水利资源、数据资源等，本申请实施例中对资源的具体形式以及内容并不作出过多限定。
待处理参数具体代表的资源元素可以根据目标资源的处理目标确定。例如，当目标资源的处理目标可以为不同地区设置的电力负载容量，以使得电网总耗能最少，此时，不同地区的电力负载容量即可以为待处理参数，处理目标即可以为电网总耗能的计算函数。目标参数可以为获得的电网总耗能最少的情况下，各个地区的电力负载容量。按照 待处理参数在目标参数的取值，可以生成目标资源的处理信息，也即，可以按照处理参数在目标参数的取值，生成各个地区的电力负载容量的提示信息或者设置指令，通过设置指令可以按照各个地区的电力负载容量进行容量设置。提示信息可以为用户展示，以供用户按照提示信息中提示的各个地区的电力负载容量对各个地区进行容量设置。
在电子商务领域中，也会涉及到参数优化问题。以较为常见的产品推荐为例，由于用户的消费习惯、关注领域、历史浏览行为等浏览特征的不同，为用户推荐的内容或者产品也不同。在实际应用中，为了提高用户的点击率，可以将用户的消费习惯、关注领域等浏览特征进行参数化，生成不同的浏览参数，并通过对多个浏览参数的设置实现对用户的点击目标的特性进行准确分析，从而查找到与用户关注度更高的目标产品。对多个浏览参数进行参数采样，并进行参数试验，以确定用户的点击概率的方案可以适用于本申请实施例的技术方案，以提高试验效率。
因此，作为一个实施例，检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法包括：
接收针对目标用户发起的浏览操作生成的浏览参数的参数优化请求；
确定与目标用户的浏览目标相匹配的参数采样算法；
可选地，响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本包括：
响应于目标用户针对浏览参数发起的样本获取请求，调用参数采样算法生成浏览参数的测试样本；
为目标用户输出测试样本的仿真结果包括：
为目标用户输出浏览参数的测试样本对应的仿真结果，以供目标用户按照测试样本配置浏览参数。
如图5所示，为本申请实施例提供的一种信息处理方法的又一个实施例的流程图，该方法可以包括：
501：检测目标用户针对参数采样算法发起的参数优化请求。
本申请实施例的技术方案可以应用于用户端，用户端例如可以为手机终端、计算机、笔记本、平板电脑、虚拟现实/增强现实设备、物联网(IoT，Internet of Things)终端等电子设备，本申请实施例中对电子设备的具体类型并不作出过多限定。
关于用户端所执行的具体内容以及技术效果，已在前述实施例中详细描述，在此不再赘述。
502：发送参数优化请求至服务端，以供服务端确定与目标用户相匹配的参数采样算法；
503：发送目标用户发起的样本获取请求至服务端，以供服务端响应样本获取请求，调用参数采样算法生成测试样本。
504：基于预设目标函数，确定测试样本的仿真结果。
505：为目标用户输出测试样本的仿真结果。
可选地，目标函数可以属于黑盒算法，此时参数优化请求即为黑盒优化请求。
本申请实施例中，用户端可以检测目标用户针对参数采样算法发起的参数优化请求。从而发送参数优化请求至服务端，服务端接收到用户端发送的参数优化请求时，可以确定该目标用户相对应的参数采样算法。之后，用户端还可以发送目标用户发起的样本获取请求至服务端，服务端接收到样本获取请求时，可以调用参数采样算法生成测试样本。并基于预设目标函数，确定测试样本的仿真结果，为目标用户输出该测试样本的仿真结果。用户端为目标用户提供样本参数的生成以及样本参数的仿真工作，通过与目标用户的交互，实现参数优化服务，可以使得目标用户可以随时选择参数采样算法进行参数优化，提高黑盒算法的利用效率。
可选地，基于预设目标函数，确定测试样本的仿真结果可以包括：接收服务端发送的测试样本，将测试样本输入目标函数计算获得测试样本的仿真结果。
可选地，基于预设目标函数，确定测试样本的仿真结果可以包括：接收服务端发送的测试样本的仿真结果，其中，该仿真结果将测试样本输入仿真函数计算获得。
作为一个实施例，基于预设目标函数，确定测试样本的仿真结果可以包括：
接收服务端发送的测试样本；
将测试样本输入目标函数，计算获得测试样本对应的计算结果。
在一种可能的设计中，接收服务端发送的测试样本可以包括：
接收服务端发送的多个测试样本；其中，多个测试样本为服务端以批处理方式调用参数采样算法生成的；
基于预设目标函数，确定测试样本的仿真结果可以包括：
将多个测试样本分别输入目标函数，计算获得多个测试样本分别对应的计算结果。
作为又一个实施例，基于预设目标函数，确定测试样本的仿真结果包括：
接收服务端发送的测试样本对应的仿真结果；其中，仿真结果为服务端将测试样本输入预设的目标函数计算获得的。
在某些实施例中，基于预设目标函数，确定测试样本的仿真结果之后，该方法还可以包括：
获取测试样本是否满足收敛条件的判断结果；
如果判断结果为测试样本满足收敛条件，则基于测试样本，确定目标样本；
如果判断结果为测试样本不满足收敛条件，则返回至发送目标用户发起的样本获取请求至服务端，以供服务端响应样本获取请求，调用参数采样算法生成测试样本继续执行。
可选地，基于测试样本，确定目标样本可以包括：
获取该测试样本之前的N次样本获取请求分别对应的历史测试样本，并基于该测试样本以及N个历史测试样本分别对应的仿真结果，从该测试样本以及N个历史测试样本中选择仿真结果最高的目标样本。
仿真结果可以为一结果指标数据，结果指标数据越高，仿真结果越好，结果指标数据越小，仿真结果越差。
需要说明的是，本申请实施例中部分步骤，在用户端执行或者在服务端执行时，执行的具体步骤相同，各个步骤的具体实现方式以及技术效果已在前述实施例中详细描述，在此不再赘述。
如图6所示，为本申请实施例提供的一种信息处理方法的又一个实施例的流程图，该方法可以包括：
601：响应于调用信息处理接口的请求，确定信息处理接口对应的处理资源。
利用信息处理接口对应的处理资源执行如下步骤：
602：检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法。
603：响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
604：基于预设目标函数，确定测试样本的仿真结果。
605：为目标用户输出测试样本的仿真结果。
本申请实施例中的信息处理接口对应的处理资源所执行的具体步骤与图1～图4所示的信息处理方法所执行的处理步骤相同，各个技术特征的具体实现方式以及技术效果已在图1～图4所示实施例中详细描述，在此不再赘述。
在实际应用中，目标用户的用户端，例如可以为手机终端、计算机、笔记本、平板电脑、虚拟现实/增强现实设备、物联网(IoT，Internet of Things)终端等电子设备，本申请实施例中对电子设备的具体类型并不作出过多限定。用户端可以与用户进行交互， 用户端可以与能够对参数进行优化的服务器进行通信。
为了便于理解，需要优化的参数为机器学习模型的网络深度、迭代次数以及每层神经元的个数构成的超参数为例，云服务端提供信息处理方法，通过与用户端进行交互为例详细介绍本申请实施例的一个应用示例。
参考图7，以用户端为手机终端M1、服务器为云服务器M2为例。手机终端M1可以检测701用户针对机器学习模型的网络深度、迭代次数以及每层神经元的个数构成的超参数发起的参数优化请求。手机终端M1可以参数优化请求发送702至云服务器M2。云服务器M2可以在检测到参数优化请求，确定703与目标用户相匹配的参数采样算法。
之后，手机终端M2可以检测704目标用户发起的样本获取请求，并将样本获取请求发送705至云服务器M2。
之后，云服务器M2可以响应于目标用户发起的样本获取请求，调用参数采样算法生成706测试样本。云服务器还可以基于预设目标函数，确定707测试样本的仿真结果；并为708目标用户输出该测试样本的仿真结果。
可选地，云服务器M2为目标用户输出测试样本的仿真结果可以包括：将测试样本的仿真结果生成7081提示页面，发送7082该提示页面至手机终端M1。手机终端M1接收到提示页面之后，可以在显示屏幕上显示7083该提示页面。
此外，测试样本的仿真结果还可以以数据字符串、短消息信息或者即时通讯消息等方式输出，本申请实施例对仿真结果的具体输出方式并不作出过多限定。
本申请实施例的技术方案可以应用于人工智能交互、数据检索、内容推荐、点击率预测、智能工厂、工业控制等多种领域，特别是在内容推荐领域，例如电子商务领域、视频直播领域、社交领域、在线教育领域中的内容推荐，以及资源分配领域，例如金融产品配置、电力资源、水利资源、供应链分配等领域中的适用性更强。
为了便于理解，以如下几个实际领域场景中的问题案例为例对本申请实施例进行详细介绍。
(1)电商领域。在电商领域中特征搜索、直播场景下的产品推荐、内容推荐和计算广告点击率等应用场景最为常见，本实施例以内容推荐场景为例，进行了实例部署。推荐场景下的通用推荐过程可以为，对选定场景的要素进行参数化设置，获得对场景产生影响的参数，利用参数标识场景的不同特征。服务端可以为目标用户展示多种黑盒算法，并对各个黑盒算法的功能以及效果进行详细描述。目标用户可以从多个黑盒算法中选择参数采样算法。并通过用户端向服务端发起参数优化请求。服务端获得目标用户发起的 参数优化请求时，可以确定目标用户选择的参数采样算法。
参数的优化过程中，需要对参数进行多次试验，以获得的目标参数。在任一次参数试验过程中，目标用户可以通过用户端发起与推荐场景相匹配的参数对应的样本获取请求，此时，服务端可以响应于目标用户发起的样本获得请求，并调用参数采样算法，生成测试样本。之后，基于预设目标函数，确定该测试样本的仿真结果。服务端确定仿真结果之后，可以将仿真结果发送至用户端，由用户端为目标用户展示测试样本的仿真结果。目标用户可以通过查看仿真结果，判断测试样本是否可以使用，如果测试样本被目标用户判断满足使用条件，目标用户即可以利用测试样本对推荐要素进行设置，并按照推荐要素对用户进行内容推荐。
以点击词推荐场景为例，用户点击APP(Application，应用程序)中的搜索框时，本系统会为该用户推荐部分搜索词(Query词)。为用户推荐搜索词的目的是挖掘用户潜在的购买需求，增加用户的使用粘性并且提高总的商品成交数量。搜索系统中利用如下的架构，结合了一种深度学习Encode-Decode(编码器—解码器)网络，也即目标网络，对搜索词的推荐进行预测。假设对搜索词的数量进行目标参数的选择。现有技术中，搜索词数量所构成的待优化参数的参数值是按照人工经验手动设置的。利用本专利的数据处理方法，可以按照以上参数优化过程自动对搜索词数量进行参数试验，服务端获得的测试样本即为搜索词数量，测试样本的仿真结果即为按照该搜索词数量进行设置之后获得的点击率或者成交率。
(2)社交领域，在社交领域中，对社交用户进行内容推荐、对学生进行素材推荐也较为常见。社交领域的推荐通常是，社交用户浏览社交应用程序，应用程序的显示界面中展示用户感兴趣的社交内容。通常，社交领域的推荐通常是，以用户的历史浏览行为、关注领域、用户信息等选项构成特征参数，不同选项的组合可以构成待处理参数。在待处理参数对应的测试样本确定时，即可以按照该测试样本进行社交内容的推荐。为了查找到社交用户感兴趣的内容，可以对参数的数量以及种类进行优化，以获得准确的社交用户内容进行测试。
本申请实施例的技术方案可以配置于云服务器中，参数优化请求可以由运维人员发起，运维人员可以将用户信息、社交类型等作为待处理参数，并利用参数采样算法不断产生测试样本。通过对测试样本的使用效果进行仿真，以获得测试参数的仿真结果，以从多个测试样本中选择目标样本。并按照目标样本对用于社交内容推荐的特征参数进行设置，实现准确推荐。
(3)金融领域。股票的指数模拟是非常重要的问题，基于线性回归，SVM(support vector machines，支持向量积)和LSTM(Long Short-Term Memory，长短记忆网络)等模型对股票指数的模拟问题较为常见，在模型使用之前，需要先建立一个合适的模型。其中，模型训练过程中可以涉及众多超参数，例如LSTM中的time_step(时间步长)，feature_dim(特征维度),hidden featrue(隐藏特征)等，而且在还涉及市场的宏观因素，微观因素，突发事件等上下文特征，这些上下文特征会影响参数的选择。利用本专利技术，可以对金融领域中需要进行设置的超参数进行测试样本的生成，通过对测试样本进行仿真，获得相应的仿真结果。
将仿真结果输出给目标用户查看，以供目标用户在从很多测试样本中选择仿真结果较好的目标样本时，利用目标样本构建指数模拟问题对应的机器学习模型，进行模型训练，获得模型参数。然后利用训练获得的机器学习模型对指数模拟问题进行实际股指的RMSE(Root Mean Squared Error，均方根误差)差异值等数据进行模拟计算。
(4)资源分配领域，以电力资源分配为例。电力资源的分配通常会涉及很多区域，每个区域可以使用相应的参数代表，这些参数分别可以分配一定比例的资源，资源的分配会影响区域经济、人口、环境等信息。
本申请实施例的技术方案可以应用于电力市场动态定价问题、电力经济负荷分配问题中。下面，主要对电力系统的具体应用领域作出详细描述。
在电力市场动态定价问题中，用户类型以及用电量的多少对电力市场的影响较为关键，可以将用户类型以及用电量等参数作为待处理参数发起参数优化请求。响应于参数优化请求，确定参数采样算法之后，可以利用参数采样算法产生待处理参数的测试样本。候选参数的设置对电力系统中收益/成本可以作为最终的优化目标，以确定该优化目标对应的目标函数，利用目标函数可以对测试样本的使用结果进行仿真，以获得仿真结果，该仿真结果即代表了系统的收益/成本的取值。之后，通过不断生成测试样本，并对测试样本的仿真，可以从很多测试样本中选择目标样本。
在电力经济负荷分配问题中，电力供应方可以同时向多个地区提供电力资源，每个地区的电力负载容量可以作为候选参数，电网总耗能可以作为目标函数的输出。利用本申请实施例的技术方案，目标用户可以对各个地区各自的电力负载容量作为待处理参数发起参数优化请求。在确定与目标用户的参数优化请求相匹配的参数采样算法之后，可以接收目标用户发起的样本获取请求，以调用参数采样算法不断审查各地区的电力负载容量的测试值对应的测试样本。在电力负荷分配领域中，目标函数可以为在电力负载容 量与电网总耗能之间存在的非线性约束关系，可以采用将目标函数对测试样本进行仿真，获得仿真结果。之后，通过不断对各个地区的电力负载容量的测试样本进行仿真，从中选择最高的仿真结果，该最高的仿真结果对应的测试样本即可以用于生成负载容量的分配策略。
为了便于理解，以电子商务领域为例，对本申请实施例的技术方案进行详细介绍。如图8所示，为本申请实施例提供的一种信息处理方法的一个实施例的流程图，该方法可以包括以下几个步骤：
801：接收目标用户针对目标资源的待处理参数发起的参数优化请求。
802：确定与目标资源的处理目标相匹配的参数采样算法。
803：响应于目标用户针对待处理参数发起的样本获取请求，调用参数采样算法生成待处理参数的测试样本。
804：基于预设目标函数，确定测试样本的仿真结果。
805：为目标用户输出待处理参数的测试样本对应仿真结果，以供目标用户按照测试样本处理目标资源。
在电力资源、水利资源的分配过程中，可以将电力资源或者水利资源在各个地区的分配结果作为一待处理参数发起参数优化请求，待处理参数具体可以为各个区域所对应的资源量，例如，电力场景中可以为区域的负载容量。
可选地，目标资源可以包括电力资源、水利资源、数据资源等，本申请实施例中对资源的具体形式以及内容并不作出过多限定。
待处理参数具体代表的资源元素可以根据目标资源的处理目标确定。例如，当目标资源的处理目标可以为不同地区设置的电力负载容量，以使得电网总耗能最少，此时，不同地区的电力负载容量即可以为待处理参数，处理目标即可以为电网总耗能的计算函数。目标参数可以为获得的电网总耗能最少的情况下，各个地区的电力负载容量。按照待处理参数在目标参数的取值，可以生成目标资源的处理信息，也即，可以按照处理参数在目标参数的取值，生成各个地区的电力负载容量的提示信息或者设置指令，通过设置指令可以按照各个地区的电力负载容量进行容量设置。提示信息可以为用户展示，以供用户按照提示信息中提示的各个地区的电力负载容量对各个地区进行容量设置。
为了便于理解，以资源处理场景为例对本申请实施例的技术方案进行详细介绍。如图9所示，为本申请实施例提供的一种信息处理方法的一个实施例的流程图，该方法可以包括以下几个步骤：
901：接收目标用户发起的浏览操作对应的浏览参数的参数优化请求；
902：确定与目标用户的浏览目标相匹配的参数采样算法。
903：响应于目标用户针对浏览参数发起的样本获取请求，调用参数采样算法生成浏览参数的测试样本。
904：基于预设目标函数，确定测试样本的仿真结果。
905：为目标用户输出浏览参数的测试样本对应的仿真结果，以供目标用户按照测试样本配置浏览参数。
在电子商务领域中，也会涉及到参数优化问题。以较为常见的产品推荐为例，由于用户的消费习惯、关注领域、历史浏览行为等浏览特征的不同，为用户推荐的内容或者产品也不同。在实际应用中，为了提高用户的点击率，可以将用户的消费习惯、关注领域等浏览特征进行参数化，生成不同的浏览参数，并通过对多个浏览参数的设置实现对用户的点击目标的特性进行准确分析，从而查找到与用户关注度更高的目标产品。对多个浏览参数进行参数采样，并进行参数试验，以确定用户的点击概率的方案可以适用于本申请实施例的技术方案，以提高试验效率。
如图10所示，为本申请实施例提供的一种信息处理装置的一个实施例的结构示意图，该装置可以包括：
请求检测模块1001：用于检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法。
第一响应模块1002：用于响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本。
第一仿真模块1003：用于基于预设目标函数，确定测试样本的仿真结果。
第一输出模块1004：用于为目标用户输出测试样本的仿真结果。
作为一个实施例，第一仿真模块可以包括：
第一发送单元，用于发送测试样本至目标用户的用户端，以供用户端将测试样本输入预设的目标函数，计算获得测试样本对应的仿真结果；
第一接收单元，用于接收用户端发送的测试样本对应的仿真结果。
作为又一个实施例，第一仿真模块可以包括：
第一计算单元，用于将测试样本输入预设的目标函数，计算获得测试样本的仿真结果。
在某些实施例中，该装置还可以包括：
节点确定单元，用于确定启动参数采样算法的目标节点；
第一响应模块可以包括：
第一响应单元，用于响应于样本获取请求，将样本获取请求发送至启动参数采样算法的目标节点；
样本获取单元，用于通过目标节点获取参数采样算法生成的测试样本。
在一种可能的设计中，第一计算单元可以包括：
模块建立单元，用于为目标函数建立仿真模块；
模块计算单元，用于将测试样本输入仿真模块，计算获得测试样本的仿真结果。
作为又一个实施例，该装置还可以包括：
标识接收模块，用于接收目标用户发送的仿真试验标识；
标识标记模块，用于基于仿真试验标识，为目标用户存储测试样本的仿真结果。
在某些实施例中，请求检测模块可以包括：
请求接收单元，用于接收目标用户发送的参数优化请求；其中，参数优化请求包括针对参数采样算法的请求服务信息；
第一存储单元，用于检测数据存储组件存在请求服务信息的存储操作时，确定目标用户在请求服务信息中请求服务的参数采样算法。
作为一种可能的实现方式，该装置还可以包括：
第一校验单元，用于基于请求服务信息中的校验信息，对目标用户的参数优化请求进行校验，获得校验结果；
第二存储单元，用于若校验结果为校验成功，则将请求服务信息存储至数据存储组件。
在某些实施例中，第一存储单元具体可以用于：
通过资源控制器获取数据存储组件在存储请求服务信息时，发起的存储通知消息；
基于存储通知消息，确定目标用户在请求服务信息中请求服务的参数采样算法。
作为一个实施例，该装置还可以包括：
收敛判断模块，用于获取测试样本是否满足预设收敛条件的判断结果；
第一处理模块，用于如果判断结果为测试样本满足收敛条件，则基于测试样本，确定目标样本；
第二处理模块，用于如果判断结果为测试样本不满足收敛条件，则跳转至第一响应模块继续执行。
在一种可能的设计中，第一响应模块可以包括：
第二响应单元，用于响应于目标用户发起的样本获取请求，以批处理方式调用参数采样算法生成多个测试样本；
第一仿真模块可以包括：
第一仿真单元，用于基于预设目标函数，确定多个测试样本分别对应的仿真结果；
第一输出模块可以包括：
第一输出单元，用于为目标用户输出多个测试样本分别对应的仿真结果。
在又一种可能的设计中，第一仿真模块可以包括：
仿真建立单元，用于为目标函数建立多个仿真模块；
仿真匹配模块，从多个仿真模块中，分别为多个测试样本匹配目标仿真模块；
第二仿真模块，用于将任一个测试样本输入测试样本对应的目标仿真模块，计算获得测试样本的仿真结果，以获得多个测试样本分别对应的仿真结果。
进一步，可选地，第一仿真单元具体可以用于：
发送多个测试样本至目标用户的用户端，以供用户端将多个测试样本分别输入预设的目标函数，计算获得多个测试样本分别对应的仿真结果；接收用户端发送的多个测试样本分别对应的仿真结果。
作为一个实施例，请求检测模块可以包括：
第二接收单元，用于接收目标用户针对目标资源的待处理参数发起的参数优化请求。
第一确定单元，用于确定与目标资源的处理目标相匹配的参数采样算法；
第一响应模块可以包括：
请求响应一单元，用于响应于目标用户针对待处理参数发起的样本获取请求，调用参数采样算法生成待处理参数的测试样本；
第一输出模块可以包括：
第二输出单元，用于为目标用户输出待处理参数的测试样本对应仿真结果，以供目标用户按照测试样本处理目标资源。
作为又一个实施例，请求检测模块可以包括：
第三接收单元，用于接收针对目标用户发起的浏览操作生成的浏览参数的参数优化请求；
第二确定单元，用于确定与目标用户的浏览目标相匹配的参数采样算法；
第一响应模块可以包括：
请求响应二单元，用于响应于目标用户针对浏览参数发起的样本获取请求，调用参数采样算法生成浏览参数的测试样本；
第一输出模块可以包括：
第三输出单元，用于为目标用户输出浏览参数的测试样本对应的仿真结果，以供目标用户按照测试样本配置浏览参数。
在某些实施例中，目标函数属于黑盒算法；请求检测模块具体可以用于：
检测所述目标用户发起的黑盒优化请求，确定与所述目标用户相匹配的参数采样算法。
图10的信息处理装置可以执行图1所示实施例的信息处理方法，其实现原理和技术效果不再赘述。对于上述实施例中的处理组件所执行的各个步骤的具体方式已经在有关该方法的实施例中进行了详细描述，此处将不做详细阐述说明。
在实际应用中，图10所示的信息处理装置可以配置为一服务器。参考图11，为本申请实施例提供的一种服务器的一个实施例的结构示意图，该服务器可以包括：存储组件1101以及处理组件1102；存储组件用于存储一条或多条计算机指令；一条或多条计算机指令被处理组件调用；
处理组件1102可以用于：
检测目标用户发起的参数优化请求，确定与目标用户相匹配的参数采样算法；响应于目标用户发起的样本获取请求，调用参数采样算法生成测试样本；基于预设目标函数，确定测试样本的仿真结果；为目标用户输出测试样本的仿真结果。
此外，处理组件还可以执行前述实施例中任一个实施例所示的信息处理方法，为了描述的简洁性考虑，在此不再赘述。本申请实施例中的服务器可以为前述实施例的服务端，可以与用户端对应用户设备进行数据或者信息的交互。
其中，处理组件1102可以包括一个或多个处理器来执行计算机指令，以完成上述的方法中的全部或部分步骤。当然处理组件也可以为一个或多个应用专用集成电路(ASIC)、数字信号处理器(DSP)、数字信号处理设备(DSPD)、可编程逻辑器件(PLD)、现场可编程门阵列(FPGA)、控制器、微控制器、微处理器或其他电子元件实现，用于执行上述信息处理方法。
存储组件1101被配置为存储各种类型的数据以支持在终端的操作。存储组件可以由任何类型的易失性或非易失性存储设备或者它们的组合实现，如静态随机存取存储器(SRAM)，电可擦除可编程只读存储器(EEPROM)，可擦除可编程只读存储器(EPROM)， 可编程只读存储器(PROM)，只读存储器(ROM)，磁存储器，快闪存储器，磁盘或光盘。
当然，计算设备必然还可以包括其他部件，例如输入/输出接口、通信组件等。输入/输出接口为处理组件和外围接口模块之间提供接口，上述外围接口模块可以是输出设备、输入设备等。通信组件被配置为便于计算设备和其他设备之间有线或无线方式的通信等。
此外，本申请实施例还提供一种计算机可读存储介质，存储介质可以存储一条或多条计算机指令，一条或多条计算机指令执行时用以实现本申请实施例中图1～图5任一种信息处理方法。
如图12所示，为本申请实施例提供的一种信息处理装置的一个实施例的结构示意图，该装置可以包括：
请求发起模块1201：用于检测目标用户针对参数采样算法发起的参数优化请求。
第一发送模块1202：用于发送目标用户提供的参数优化请求至服务端，以供服务端确定与目标用户相匹配的参数采样算法；
第二发送模块1203：用于发送目标用户发起的样本获取请求至服务端，以供服务端响应样本获取请求，调用参数采样算法生成测试样本。
第二仿真模块1204：用于基于预设目标函数，确定测试样本的仿真结果。
第二输出模块1205：用于为目标用户输出测试样本的仿真结果。
作为一个实施例，第二仿真模块可以包括：
样本接收单元，用于接收服务端发送的测试样本；
样本仿真单元，用于将测试样本输入目标函数，计算获得测试样本对应的计算结果。
在某些实施例中，样本接收单元具体可以用于：接收服务端发送的多个测试样本；其中，多个测试样本为服务端以批处理方式调用参数采样算法生成的。
样本仿真单元具体可以用于：将多个测试样本分别输入目标函数，计算获得多个测试样本分别对应的计算结果。
作为又一个实施例，第二仿真模块可以包括：
结果接收单元，用于接收服务端发送的测试样本对应的仿真结果；其中，仿真结果为服务端将测试样本输入预设的目标函数计算获得的。
在某些实施例中，该装置还可以包括：
收敛判断模块，用于获取测试样本是否满足预设收敛条件的判断结果；
第一处理模块，用于如果判断结果为测试样本满足收敛条件，则基于测试样本，确定目标样本；
第二处理模块，用于如果判断结果为测试样本不满足收敛条件，则跳转至第二发送模块继续执行。
图12的信息处理装置可以执行图7所示实施例的信息处理方法，其实现原理和技术效果不再赘述。对于上述实施例中的处理组件所执行的各个步骤的具体方式已经在有关该方法的实施例中进行了详细描述，此处将不做详细阐述说明。
在实际应用中，图12所示的信息处理装置可以配置为一用户设备。参考图13，为本申请实施例提供的一种用户设备的一个实施例的结构示意图，该服务器可以包括：存储组件1301以及处理组件1302；存储组件1301用于存储一条或多条计算机指令；一条或多条计算机指令被处理组件1302调用。
处理组件1302可以用于：
检测目标用户针对参数采样算法发起的参数优化请求；发送目标用户提供的参数优化请求至服务端，以供服务端确定与目标用户相匹配的参数采样算法；发送目标用户发起的样本获取请求至服务端，以供服务端响应样本获取请求，调用参数采样算法生成测试样本；基于预设目标函数，确定测试样本的仿真结果；为目标用户输出测试样本的仿真结果。
本申请实施例中所示的用户设备具体可以作为前述实施例中的用户端，关于本申请实施例中各个步骤的具体处理方法以及有益效果可以参考前述实施例中用户端的相关描述，在此不再赘述。
此外，处理组件还可以执行前述实施例中任一个实施例所示的信息处理方法，为了描述的简洁性考虑，在此不再赘述。
其中，处理组件1302可以包括一个或多个处理器来执行计算机指令，以完成上述的方法中的全部或部分步骤。当然处理组件也可以为一个或多个应用专用集成电路(ASIC)、数字信号处理器(DSP)、数字信号处理设备(DSPD)、可编程逻辑器件(PLD)、现场可编程门阵列(FPGA)、控制器、微控制器、微处理器或其他电子元件实现，用于执行上述信息处理方法。
存储组件1301被配置为存储各种类型的数据以支持在终端的操作。存储组件可以由任何类型的易失性或非易失性存储设备或者它们的组合实现，如静态随机存取存储器(SRAM)，电可擦除可编程只读存储器(EEPROM)，可擦除可编程只读存储器(EPROM)， 可编程只读存储器(PROM)，只读存储器(ROM)，磁存储器，快闪存储器，磁盘或光盘。
当然，计算设备必然还可以包括其他部件，例如输入/输出接口、通信组件等。输入/输出接口为处理组件和外围接口模块之间提供接口，上述外围接口模块可以是输出设备、输入设备等。通信组件被配置为便于计算设备和其他设备之间有线或无线方式的通信等。
此外，本申请实施例还提供一种计算机可读存储介质，存储介质可以存储一条或多条计算机指令，一条或多条计算机指令执行时用以实现本申请实施例中图6所示的信息处理方法。
以上所描述的装置实施例仅仅是示意性的，其中作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到各实施方式可借助加必需的通用硬件平台的方式来实现，当然也可以通过硬件和软件结合的方式来实现。基于这样的理解，上述技术方案本质上或者说对现有技术做出贡献的部分可以以计算机产品的形式体现出来，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (22)

  1. 一种信息处理方法，其特征在于，包括：
    检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
    响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
    基于预设目标函数，确定所述测试样本的仿真结果；
    为所述目标用户输出所述测试样本的仿真结果。
  2. 根据权利要求1所述的方法，其特征在于，所述基于预设目标函数，确定所述测试样本的仿真结果包括：
    发送所述测试样本至所述目标用户的用户端，以供所述用户端将所述测试样本输入预设的所述目标函数，计算获得所述测试样本对应的仿真结果；
    接收所述用户端发送的所述测试样本对应的仿真结果。
  3. 根据权利要求1所述的方法，其特征在于，所述检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法之后，还包括：
    确定启动所述参数采样算法的目标节点；
    所述响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本包括：
    响应于所述样本获取请求，将所述样本获取请求发送至启动所述参数采样算法的目标节点；
    通过所述目标节点获取所述参数采样算法生成的测试样本。
  4. 根据权利要求1所述的方法，其特征在于，所述基于预设目标函数，确定所述测试样本的仿真结果包括：
    将所述测试样本输入预设的所述目标函数，计算获得所述测试样本的仿真结果。
  5. 根据权利要求4所述的方法，其特征在于，所述将所述测试样本输入预设的所述目标函数，计算获得所述测试样本的仿真结果包括：
    为所述目标函数建立仿真模块；
    将所述测试样本输入所述仿真模块，计算获得所述测试样本的仿真结果。
  6. 根据权利要求1所述的方法，其特征在于，还包括：
    接收所述目标用户发送的仿真试验标识；
    基于所述仿真试验标识，为所述目标用户存储所述测试样本的仿真结果。
  7. 根据权利要求1所述的方法，其特征在于，所述检测目标用户发起的参数优化 请求，确定与所述目标用户相匹配的参数采样算法包括：
    接收目标用户发送的参数优化请求；其中，所述参数优化请求包括针对参数采样算法的请求服务信息；
    检测数据存储组件存在所述请求服务信息的存储操作时，确定所述目标用户在所述请求服务信息中请求服务的参数采样算法。
  8. 根据权利要求7所述的方法，其特征在于，所述请求服务信息还包括校验信息；所述接收目标用户发送的参数优化请求之后，还包括：
    基于所述请求服务信息中的校验信息，对所述目标用户的参数优化请求进行校验，获得校验结果；
    若所述校验结果为校验成功，则将所述请求服务信息存储至所述数据存储组件。
  9. 根据权利要求7所述的方法，其特征在于，所述检测数据存储组件存在所述请求服务信息的存储操作时，确定所述目标用户在所述请求服务信息中请求服务的参数采样算法包括：
    通过资源控制器获取所述数据存储组件在存储所述请求服务信息时，发起的存储通知消息；
    基于所述存储通知消息，确定所述目标用户在所述请求服务信息中请求服务的参数采样算法。
  10. 根据权利要求1所述的方法，其特征在于，所述基于预设目标函数，确定所述测试样本的仿真结果之后，该方法还包括：
    获取所述测试样本是否满足预设收敛条件的判断结果；
    如果所述判断结果为所述测试样本满足收敛条件，则基于所述测试样本，确定目标样本；
    如果所述判断结果为所述测试样本不满足收敛条件，则返回至所述响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本继续执行。
  11. 根据权利要求1所述的方法，其特征在于，所述目标函数属于黑盒算法；所述检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法包括：
    检测所述目标用户发起的黑盒优化请求，确定与所述目标用户相匹配的参数采样算法。
  12. 一种信息处理方法，其特征在于，还包括：
    接收目标用户针对目标资源的待处理参数发起的参数优化请求；
    确定与所述目标资源的处理目标相匹配的参数采样算法；
    响应于所述目标用户针对所述待处理参数发起的样本获取请求，调用所述参数采样算法生成所述待处理参数的测试样本；
    基于预设目标函数，确定所述测试样本的仿真结果；
    为所述目标用户输出所述待处理参数的测试样本对应仿真结果，以供所述目标用户按照所述测试样本处理所述目标资源。
  13. 一种信息处理方法，其特征在于，包括：
    接收目标用户发起的浏览操作对应的浏览参数的参数优化请求；
    确定与所述目标用户的浏览目标相匹配的参数采样算法；
    响应于所述目标用户针对所述浏览参数发起的样本获取请求，调用所述参数采样算法生成所述浏览参数的测试样本；
    基于预设目标函数，确定所述测试样本的仿真结果；
    为所述目标用户输出所述浏览参数的测试样本对应的仿真结果，以供所述目标用户按照所述测试样本配置所述浏览参数。
  14. 一种信息处理方法，其特征在于，包括：
    检测目标用户针对参数采样算法发起的参数优化请求；
    发送所述参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；
    发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；
    基于预设目标函数，确定所述测试样本的仿真结果；
    为所述目标用户输出所述测试样本的仿真结果。
  15. 根据权利要求14所述的方法，其特征在于，所述基于预设目标函数，确定所述测试样本的仿真结果包括：
    接收所述服务端发送的测试样本；
    将所述测试样本输入所述目标函数，计算获得所述测试样本对应的计算结果。
  16. 根据权利要求15所述的方法，其特征在于，所述接收所述服务端发送的所述测试样本包括：
    接收所述服务端发送的多个测试样本；其中，所述多个测试样本为所述服务端以批处理方式调用所述参数采样算法生成的；
    所述基于预设目标函数，确定所述测试样本的仿真结果包括：
    将所述多个测试样本分别输入所述目标函数，计算获得所述多个测试样本分别对应的计算结果。
  17. 根据权利要求14所述的方法，其特征在于，所述基于预设目标函数，确定所述测试样本的仿真结果之后，还包括：
    获取所述测试样本是否满足预设收敛条件的判断结果；
    如果所述判断结果为所述测试样本满足收敛条件，则基于所述测试样本，确定目标样本；
    如果所述判断结果为所述测试样本不满足收敛条件，则返回至发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本继续执行。
  18. 一种信息处理方法，其特征在于，包括：
    响应于调用信息处理接口的请求，确定所述信息处理接口对应的处理资源；
    利用所述信息处理接口对应的处理资源执行如下步骤：
    检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
    响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
    基于预设目标函数，确定所述测试样本的仿真结果；
    为所述目标用户输出所述测试样本的仿真结果。
  19. 一种信息处理装置，其特征在于，包括：
    请求检测模块，用于检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；
    第一响应模块，用于响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；
    第一仿真模块，用于基于预设目标函数，确定所述测试样本的仿真结果；
    第一输出模块，用于为所述目标用户输出所述测试样本的仿真结果。
  20. 一种信息处理装置，其特征在于，包括：
    请求发起模块，用于检测目标用户针对参数采样算法发起的参数优化请求；
    第一发送模块，用于发送目标用户提供的参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；
    第二发送模块，用于发送所述目标用户发起的样本获取请求至所述服务端，以供所 述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；
    第二仿真模块，用于基于预设目标函数，确定所述测试样本的仿真结果；
    第二输出模块，用于为所述目标用户输出所述测试样本的仿真结果。
  21. 一种服务器，其特征在于，包括：存储组件以及处理组件；所述存储组件用于存储一条或多条计算机指令；所述一条或多条计算机指令被所述处理组件调用；
    所述处理组件用于：
    检测目标用户发起的参数优化请求，确定与所述目标用户相匹配的参数采样算法；响应于所述目标用户发起的样本获取请求，调用所述参数采样算法生成测试样本；基于预设目标函数，确定所述测试样本的仿真结果；为所述目标用户输出所述测试样本的仿真结果。
  22. 一种用户设备，其特征在于，包括：存储组件以及处理组件；所述存储组件用于存储一条或多条计算机指令；所述一条或多条计算机指令被所述处理组件调用；
    所述处理组件用于：
    检测目标用户针对参数采样算法发起的参数优化请求；发送目标用户提供的参数优化请求至服务端，以供所述服务端确定与所述目标用户相匹配的参数采样算法；发送所述目标用户发起的样本获取请求至所述服务端，以供所述服务端响应所述样本获取请求，调用所述参数采样算法生成测试样本；基于预设目标函数，确定所述测试样本的仿真结果；为所述目标用户输出所述测试样本的仿真结果。
PCT/CN2022/072913 2021-01-25 2022-01-20 信息处理方法及装置、服务器及用户设备 WO2022156731A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP22742210.2A EP4283511A4 (en) 2021-01-25 2022-01-20 INFORMATION PROCESSING METHOD AND DEVICE, SERVER AND USER EQUIPMENT
US18/357,442 US20240020556A1 (en) 2021-01-25 2023-07-24 Information processing method and apparatus, server, and user device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110099097.6A CN112784418A (zh) 2021-01-25 2021-01-25 信息处理方法及装置、服务器及用户设备
CN202110099097.6 2021-01-25

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/357,442 Continuation US20240020556A1 (en) 2021-01-25 2023-07-24 Information processing method and apparatus, server, and user device

Publications (1)

Publication Number Publication Date
WO2022156731A1 true WO2022156731A1 (zh) 2022-07-28

Family

ID=75759094

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/072913 WO2022156731A1 (zh) 2021-01-25 2022-01-20 信息处理方法及装置、服务器及用户设备

Country Status (4)

Country Link
US (1) US20240020556A1 (zh)
EP (1) EP4283511A4 (zh)
CN (1) CN112784418A (zh)
WO (1) WO2022156731A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112784418A (zh) * 2021-01-25 2021-05-11 阿里巴巴集团控股有限公司 信息处理方法及装置、服务器及用户设备
CN114338820A (zh) * 2021-12-31 2022-04-12 中国电信股份有限公司 网络访问方法、装置、电子设备及存储介质
CN116884759B (zh) * 2023-07-19 2024-03-22 重庆望变电气（集团）股份有限公司 一种铁芯叠放工艺方案生成系统和方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109447277A (zh) * 2018-10-19 2019-03-08 厦门渊亭信息科技有限公司 一种通用的机器学习超参黑盒优化方法及系统
US20200097853A1 (en) * 2017-06-02 2020-03-26 Google Llc Systems and Methods for Black Box Optimization
CN111325356A (zh) * 2019-12-10 2020-06-23 四川大学 一种基于演化计算的神经网络搜索分布式训练系统及训练方法
US20200382388A1 (en) * 2019-05-31 2020-12-03 International Business Machines Corporation Constrained optimization of cloud micro services
CN112784418A (zh) * 2021-01-25 2021-05-11 阿里巴巴集团控股有限公司 信息处理方法及装置、服务器及用户设备

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US12026612B2 (en) * 2017-06-02 2024-07-02 Google Llc Optimization of parameter values for machine-learned models
CN108763064B (zh) * 2018-05-10 2020-07-07 南京大学 一种基于黑盒函数与机器学习的代码测试生成方法和装置
WO2020149971A2 (en) * 2019-01-16 2020-07-23 Google Llc Robust and data-efficient blackbox optimization
CN111859114A (zh) * 2020-06-18 2020-10-30 北京百度网讯科技有限公司 推荐系统的优化方法、装置、设备和计算机存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200097853A1 (en) * 2017-06-02 2020-03-26 Google Llc Systems and Methods for Black Box Optimization
CN109447277A (zh) * 2018-10-19 2019-03-08 厦门渊亭信息科技有限公司 一种通用的机器学习超参黑盒优化方法及系统
US20200382388A1 (en) * 2019-05-31 2020-12-03 International Business Machines Corporation Constrained optimization of cloud micro services
CN111325356A (zh) * 2019-12-10 2020-06-23 四川大学 一种基于演化计算的神经网络搜索分布式训练系统及训练方法
CN112784418A (zh) * 2021-01-25 2021-05-11 阿里巴巴集团控股有限公司 信息处理方法及装置、服务器及用户设备

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4283511A4

Also Published As

Publication number Publication date
CN112784418A (zh) 2021-05-11
US20240020556A1 (en) 2024-01-18
EP4283511A1 (en) 2023-11-29
EP4283511A4 (en) 2024-10-23

Similar Documents

Publication Publication Date Title
WO2022156731A1 (zh) 信息处理方法及装置、服务器及用户设备
Dai et al. Cloud infrastructure resource allocation for big data applications
Sun et al. Cloud service selection: State-of-the-art and future research directions
Tao et al. Modelling of combinable relationship-based composition service network and the theoretical proof of its scale-free characteristics
CN114265979B (zh) 确定融合参数的方法、信息推荐方法和模型训练方法
Li et al. A new QoS-aware web service recommendation system based on contextual feature recognition at server-side
Chen et al. Your neighbors alleviate cold-start: On geographical neighborhood influence to collaborative web service QoS prediction
Ding et al. Decision support for personalized cloud service selection through multi-attribute trustworthiness evaluation
WO2014193399A1 (en) Influence score of a brand
Skopik et al. Start trusting strangers? bootstrapping and prediction of trust
Wu et al. Reputation revision method for selecting cloud services based on prior knowledge and a market mechanism
US20230315618A1 (en) Data processing method and device, computing device, and test reduction device
Liu et al. A virtual machine migration strategy based on time series workload prediction using cloud model
WO2022174755A1 (zh) 信息处理方法及装置、计算设备
Choi et al. Quality evaluation and best service choice for cloud computing based on user preference and weights of attributes using the analytic network process
Xu et al. MeURep: A novel user reputation calculation approach in personalized cloud services
Xu et al. Novel model of e-commerce marketing based on big data analysis and processing
Li et al. SCCAF: A Secure and Compliant Continuous Assessment Framework in Cloud‐Based IoT Context
Caunedo et al. Asymmetry, complementarities, and state dependence in Federal Reserve forecasts
Zhao et al. Assessment of recommendation trust for access control in open networks
Wu Context‐Aware Cloud Service Selection Model for Mobile Cloud Computing Environments
Hosseinnezhad et al. A probabilistic trust model for cloud services using Bayesian networks
Ma et al. User preferences-aware recommendation for trustworthy cloud services based on fuzzy clustering
CN106817296B (zh) 信息推荐的测试方法、装置以及电子设备
Zhang et al. Web service composition algorithm based on TOPSIS

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22742210

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2022742210

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022742210

Country of ref document: EP

Effective date: 20230825