Data processing method and device, electronic equipment and computer readable storage medium

Abstract

The invention relates to the technical field of data processing, is applied to smart medical and financial science and technology scenes, and provides a data processing method and device, electronic equipment and a computer readable storage medium. Constructing a multi-dimensional verification set according to the business feature information and a preset hierarchical verification set construction module; performing multi-dimensional over-fitting monitoring processing on the to-be-monitored model according to the multi-dimensional verification set to obtain over-fitting monitoring information; under the condition that the over-fitting monitoring information represents that the to-be-monitored model is over-fitted, performing evaluation processing on the over-fitting monitoring information to obtain an over-fitting type and an over-fitting degree; determining an overfitting adjustment strategy according to the overfitting type and the overfitting degree; and performing dynamic adjustment processing on the to-be-monitored model based on the over-fitting adjustment strategy. According to the technical scheme, the particularity of various services can be considered, and the problem of overfitting of various models can be found and effectively relieved in time.

Description

Technical Field

[0001] The embodiments of this application relate to, but are not limited to, the field of data processing technology, and in particular to a data processing method, apparatus, electronic device, and computer-readable storage medium. Background Technology

[0002] With the continuous development of science and technology, artificial intelligence has been widely promoted and applied. For example, in the financial field, financial models can be used to manage and process various user businesses in a unified manner; in the insurance field, insurance business models can be used to manage and process various insurance businesses and insurance claims intelligently; and in the field of smart healthcare, smart healthcare models can be used to centrally manage various medical matters. However, in the process of fine-tuning models, current methods generally only focus on the single indicator of the decline in validation set accuracy, lacking consideration for the specific characteristics of various businesses, making it difficult to promptly identify and effectively alleviate the problem of overfitting in various models. Summary of the Invention

[0003] The following is an overview of the subject matter described in detail herein. This overview is not intended to limit the scope of the claims.

[0004] To address the problems mentioned in the background section, embodiments of this application provide a data processing method, apparatus, electronic device, and computer-readable storage medium that can take into account the specific characteristics of various businesses and promptly detect and effectively mitigate the problem of overfitting in various models.

[0005] In a first aspect, embodiments of this application provide a data processing method, the data processing method comprising: Obtain business characteristic information; A multidimensional verification set is constructed based on the business characteristic information and the preset hierarchical verification set construction module; Based on the multidimensional validation set, multidimensional overfitting monitoring processing is performed on the model to be monitored to obtain overfitting monitoring information; When the overfitting monitoring information indicates that the monitored model is overfitting, the overfitting monitoring information is evaluated to obtain the overfitting type and the degree of overfitting. Determine the overfitting adjustment strategy based on the overfitting type and the degree of overfitting; The overfitting adjustment strategy is used to dynamically adjust the model to be monitored.

[0006] Secondly, embodiments of this application also provide a data processing apparatus, the data processing apparatus comprising: The acquisition unit is used to acquire business characteristic information; The construction unit is used to construct a multidimensional verification set based on the business feature information and the preset hierarchical verification set construction module; The monitoring unit is used to perform multi-dimensional overfitting monitoring processing on the model to be monitored based on the multi-dimensional validation set, and obtain overfitting monitoring information. An evaluation unit is used to evaluate the overfitting monitoring information when the overfitting monitoring information indicates that the monitored model is overfitting, and to obtain the overfitting type and the degree of overfitting. A combination unit is used to determine an overfitting adjustment strategy based on the overfitting type and the degree of overfitting. The adjustment unit is used to dynamically adjust the model to be monitored based on the overfitting adjustment strategy.

[0007] Thirdly, embodiments of this application also provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data processing method described in the first aspect above.

[0008] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer-executable instructions for performing the data processing method described in the first aspect above.

[0009] The data processing method according to the embodiments provided in this application has at least the following beneficial effects: In the data processing process, business feature information is first obtained; then, a multi-dimensional validation set is constructed based on the business feature information and a preset hierarchical validation set construction module; next, overfitting monitoring information is obtained by performing multi-dimensional overfitting monitoring on the model to be monitored based on the multi-dimensional validation set; then, when the overfitting monitoring information indicates that the model to be monitored is overfitted, the overfitting monitoring information is evaluated to obtain the overfitting type and degree; next, an overfitting adjustment strategy is determined based on the overfitting type and degree; finally, the model to be monitored is dynamically adjusted based on the overfitting adjustment strategy. Through the above technical solution, various business-specific characteristics can be considered during the model overfitting monitoring process, and various models' overfitting problems can be detected and effectively mitigated in a timely manner. Attached Figure Description

[0010] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0011] Figure 1 This is a schematic flowchart of a data processing method provided in one embodiment of this application; Figure 2yes Figure 1 A schematic diagram of a specific implementation method of step S200; Figure 3 yes Figure 1 A schematic diagram of a specific implementation of step S300; Figure 4 yes Figure 1 A schematic diagram of a specific implementation of step S400; Figure 5 yes Figure 1 A schematic diagram of a specific implementation of step S500; Figure 6 yes Figure 1 A schematic diagram of a specific implementation method of step S600; Figure 7 Is it completed? Figure 1 A flowchart illustrating a specific implementation method following step S600; Figure 8 This is a schematic diagram of a data processing apparatus provided in one embodiment of this application; Figure 9 This is a schematic diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0013] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0014] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0015] AI is a new technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. Artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produce new intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. Artificial intelligence can simulate the information processes of human consciousness and thought. Furthermore, artificial intelligence utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results—the theories, methods, technologies, and application systems available for use.

[0016] Foundational technologies in artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0017] Artificial intelligence, or AI, is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0018] The servers involved in artificial intelligence technology can be standalone servers or cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0019] This application provides a data processing method, apparatus, electronic device, and computer-readable storage medium. In the data processing process, firstly, business characteristic information is acquired. Then, a multi-dimensional validation set is constructed based on the business characteristic information and a pre-defined hierarchical validation set construction module. Next, multi-dimensional overfitting monitoring is performed on the model to be monitored based on the multi-dimensional validation set to obtain overfitting monitoring information. Then, when the overfitting monitoring information indicates that the model to be monitored is overfitting, the overfitting monitoring information is evaluated to obtain the overfitting type and degree. Next, an overfitting adjustment strategy is determined based on the overfitting type and degree. Finally, the model to be monitored is dynamically adjusted based on the overfitting adjustment strategy. Through the above technical solution, various business-specific characteristics can be considered during the model overfitting monitoring process, enabling timely detection and effective mitigation of various model overfitting problems.

[0020] The data processing method provided in this application relates to the field of data processing technology. The data processing method provided in this application can be applied to a terminal or a server, and can also be software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0021] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0022] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0023] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0024] like Figure 1 As shown, Figure 1This is a flowchart illustrating a data processing method provided in one embodiment of this application. The data processing method includes the following steps: Step S100: Obtain business characteristic information.

[0025] The data processing method provided in this application first acquires business characteristic information during data processing, and then constructs a multi-dimensional validation set based on the business characteristic information to prepare for subsequent overfitting validation of the model to be monitored. For example, in the financial industry, a risk management model can realize business functions such as anti-fraud identification, credit approval, and credit limit calculation; an intelligent marketing model can realize business functions such as customer segmentation, precise recommendation, and outbound calling strategies; an investment research and decision-making model can realize functions such as policy research report interpretation, asset pricing, and intelligent investment advisory; an operational efficiency improvement model can realize functions such as code generation, document interpretation and translation, and knowledge retrieval; and a supply chain model can realize functions such as automated rating and access verification. Therefore, based on the business characteristic information of each of the above models, a corresponding multi-dimensional validation set can be constructed to prepare for overfitting validation of the model, making the overfitting validation more comprehensive and accurate. Alternatively, in the smart healthcare industry, clinical decision support models can achieve disease early warning, intelligent consultation and triage models can achieve smart outpatient clinics, medical record generation models can achieve intelligent generation of medical records and automatic writing of imaging reports, intelligent health management models can achieve health assessment and physical examination assessment, and medical insurance cost control models can achieve intelligent auditing. Therefore, based on the business characteristic information of the above models, corresponding multi-dimensional validation sets can be constructed to prepare for overfitting validation of the models, so that overfitting validation can be more comprehensive and accurate.

[0026] It is worth noting that a corresponding multidimensional validation set can be determined based on business characteristic information. Subsequently, the determined multidimensional validation set can be used to perform overfit validation on the corresponding model. This allows the business characteristics to be taken into account during the overfit validation process, making the overfit validation process more accurate.

[0027] It is worth noting that in the process of acquiring business characteristic information, when it involves processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent is always obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when this application embodiment needs to obtain sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after explicitly obtaining the user's separate permission or consent is the necessary user-related data for the normal operation of this application embodiment acquired.

[0028] Step S200: Construct a multidimensional verification set based on business characteristic information and a preset hierarchical verification set construction module.

[0029] The data processing method provided in this application embodiment, after obtaining business feature information during data processing, can construct a multidimensional validation set based on the business feature information and a pre-set hierarchical validation set construction module, in order to prepare for subsequent multidimensional overfitting monitoring of the model to be monitored; through the multidimensional validation set, multidimensional overfitting monitoring of the model to be monitored can be performed, making the overfitting verification of the model more comprehensive and improving its rationality and accuracy.

[0030] It's worth noting that the hierarchical validation set construction module is a functional component in machine learning projects used to create a balanced validation dataset. Its core objectives are to maintain a hierarchical data structure, prevent information leakage, and improve the reliability of model evaluation. In supervised learning scenarios, the validation set is used to tune hyperparameters and evaluate the model's generalization ability; the hierarchical validation set construction module can ensure the consistency of the validation set's distribution with the training set across key dimensions by using stratified sampling when dealing with class-imbalanced or time-dependent data.

[0031] like Figure 2 As shown, constructing a multidimensional validation set based on business characteristic information and a pre-defined hierarchical validation set construction module can include the following steps: Step S210: Determine surface diversity information, business rule information, adversarial example information, and time stratification information based on business characteristic information; Step S220: Based on surface diversity information, construct a surface diversity verification set using a hierarchical verification set construction module; and based on business rule information, construct a business rule verification set using a hierarchical verification set construction module; and based on adversarial example information, construct an adversarial example verification set using a hierarchical verification set construction module; and based on time hierarchical information, construct a time hierarchical verification set using a hierarchical verification set construction module. Step S230: The surface diversity validation set, business rule validation set, adversarial example validation set, and time-layered validation set are determined as the multidimensional validation set.

[0032] For steps S210 to S230, in the process of determining surface diversity information, business rule information, adversarial example information, and temporal stratification information based on business feature information, firstly, surface diversity information, business rule information, adversarial example information, and temporal stratification information are determined based on business feature information; then, based on surface diversity information, a surface diversity validation set is constructed using a hierarchical validation set construction module; and, based on business rule information, a business rule validation set is constructed using a hierarchical validation set construction module; and, based on adversarial example information, an adversarial example validation set is constructed using a hierarchical validation set construction module; and, based on temporal stratification information, a temporal stratification validation set is constructed using a hierarchical validation set construction module; finally, the constructed surface diversity validation set, business rule validation set, adversarial example validation set, and temporal stratification validation set are determined as a multidimensional validation set. Through the above technical solution, the subsequent overfitting monitoring process can consider surface diversity, business rules, adversarial examples, and temporal stratification, making overfitting monitoring more comprehensive, accurate, and reasonable.

[0033] For example, in the financial industry, the surface diversity validation set can include synonyms such as assessing loan risk and analyzing credit risk; the business rule validation set can include edge cases, such as applicants with no credit history but stable income; the adversarial sample validation set can include deliberately involved contradictory risk feature samples; and the time-layered validation set can include recently emerging risk pattern cases.

[0034] Step S300: Perform multi-dimensional overfit monitoring processing on the model to be monitored based on the multi-dimensional validation set to obtain overfit monitoring information.

[0035] The data processing method provided in this application embodiment, after constructing a multidimensional validation set based on business feature information and a preset hierarchical validation set construction module, can perform multidimensional overfitting monitoring processing on the model to be monitored based on the multidimensional validation set, thereby obtaining corresponding overfitting monitoring information; performing multidimensional overfitting monitoring processing on the model to be monitored based on the multidimensional validation set makes overfitting monitoring more comprehensive, reasonable and accurate.

[0036] For example, in the financial industry, the model to be monitored is an intelligent marketing model. Therefore, a multi-dimensional overfitting monitoring process can be performed on the intelligent marketing model based on a multi-dimensional validation set to obtain corresponding overfitting monitoring information. Alternatively, in the smart healthcare industry, the model to be monitored is an intelligent consultation and triage model. Therefore, a multi-dimensional overfitting monitoring process can be performed on the intelligent marketing model based on a multi-dimensional validation set to obtain corresponding overfitting monitoring information.

[0037] like Figure 3As shown, the process of performing multi-dimensional overfitting monitoring on the model to be monitored based on the multi-dimensional validation set to obtain overfitting monitoring information can include the following steps: Step S310: Based on the multidimensional validation set, use the preset instruction consistency analyzer to perform overfit monitoring processing on the model to be monitored to obtain the first monitoring information. Step S320: Based on the multidimensional validation set, use a preset answer diversity evaluator to perform overfitting monitoring on the model to be monitored to obtain the second monitoring information; Step S330: Based on the multidimensional validation set, use a preset confidence anomaly detector to perform overfitting detection processing on the model to be monitored to obtain the third monitoring information. Step S340: Based on the multidimensional validation set, use the preset business rule compliance checker to perform overfit monitoring processing on the model to be monitored to obtain the fourth monitoring information. Step S350: The first monitoring information, the second monitoring information, the third monitoring information, and the fourth monitoring information are identified as overfitting monitoring information.

[0038] For steps S310 to S350, in the process of obtaining overfitting monitoring information by performing multi-dimensional overfitting monitoring on the model to be monitored based on the multi-dimensional validation set, firstly, based on the multi-dimensional validation set, a preset instruction consistency analyzer is used to perform overfitting monitoring on the model to be monitored to obtain first monitoring information; then, based on the multi-dimensional validation set, a preset answer diversity evaluator is used to perform overfitting monitoring on the model to be monitored to obtain second monitoring information; next, based on the multi-dimensional validation set, a preset confidence anomaly detector is used to perform overfitting monitoring on the model to be monitored to obtain third monitoring information; then, based on the multi-dimensional validation set, a preset business rule compliance checker is used to perform overfitting monitoring on the model to be monitored to obtain fourth monitoring information; finally, the first, second, third, and fourth monitoring information are determined as overfitting monitoring information. Through the above technical solution, the overfitting test of the model to be monitored can be more comprehensive and accurate.

[0039] It is worth noting that the instruction consistency analyzer can monitor the consistency of the model's responses to different expressions of the same semantics. The answer diversity evaluator can quantify the innovativeness and diversity of the model's output. The confidence anomaly detector can identify anomalous differences in the model's confidence levels for similar questions. The business rule compliance checker can verify whether the output conforms to business logic. For example, in the financial field, the instruction consistency analyzer can monitor the consistency of the model's responses to different expressions of the same semantics in the financial domain; the answer diversity evaluator can quantify the innovativeness and diversity of the financial model's output; the confidence anomaly detector can identify anomalous differences in the financial model's confidence levels for similar questions; and the business rule compliance checker can verify whether the output conforms to financial business logic. Alternatively, in the medical field, the instruction consistency analyzer can monitor the consistency of the model's responses to different expressions of the same semantics in the medical domain; the answer diversity evaluator can quantify the innovativeness and diversity of the medical model's output; the confidence anomaly detector can identify anomalous differences in the medical model's confidence levels for similar questions; and the business rule compliance checker can verify whether the output conforms to medical business logic.

[0040] Step S400: When the overfitting monitoring information indicates that the monitored model is overfitting, the overfitting monitoring information is evaluated and processed to obtain the overfitting type and degree.

[0041] The data processing method provided in this application, after obtaining overfitting monitoring information by performing multi-dimensional overfitting monitoring processing on the model to be monitored based on the multi-dimensional validation set, can evaluate the overfitting monitoring information when the overfitting monitoring information indicates that the model to be monitored is overfitting, thereby obtaining the overfitting type and overfitting degree, in order to prepare for subsequent dynamic adjustment processing of the model to be monitored.

[0042] It is worth noting that overfitting monitoring information is only evaluated and processed when it indicates that the monitored model is overfitting; if it indicates that the monitored model is not overfitting, it is not evaluated and processed.

[0043] like Figure 4 As shown, evaluating and processing overfitting monitoring information to obtain the overfitting type and degree can include the following steps: Step S410: The overfitting monitoring information is segmented to obtain multiple overfitting segmentation monitoring information. Step S420: Perform keyword recognition processing on multiple overfitting segmentation monitoring information to obtain the overfitting type and overfitting degree.

[0044] For steps S410 to S420, in the process of evaluating and processing the overfitting monitoring information to obtain the overfitting type and degree, the overfitting monitoring information is first segmented to obtain multiple overfitting segmented monitoring information; then, keyword recognition processing is performed on the multiple overfitting segmented monitoring information to obtain the overfitting type and degree, providing a data basis for the subsequent dynamic adjustment of the monitoring model.

[0045] It is worth noting that segmenting the overfitting monitoring information can yield multiple overfitting segmentation monitoring information, which prepares the data for subsequent keyword identification. Performing keyword identification on multiple overfitting segmentation monitoring information can yield the overfitting type and degree, making the determination of the overfitting type and degree simpler, faster, and more accurate.

[0046] It is worth noting that overfitting can be categorized based on data characteristics into sample-level overfitting, feature-level overfitting, and distribution-level overfitting. For sample-level overfitting, a medical imaging model might mistakenly identify artifacts from a hospital's CT scan as disease markers, leading to a decrease in accuracy after switching equipment. For feature-level overfitting, a credit scoring model might use "user completion time" as a strong feature, but this feature fluctuates drastically across different channels. For distribution-level overfitting, an anti-fraud model trained on quarterly data might fail to adapt to changes in transaction patterns during holidays. Overfitting can also be categorized based on model structure into overcapacity overfitting and hierarchical depth overfitting. Furthermore, overfitting can be categorized based on the time dimension into time-series overfitting.

[0047] It is worth noting that the degree of overfitting can be represented numerically; the higher the value, the greater the degree of overfitting. For example, the value "0-1" can be used to represent the degree of overfitting, with the closer to "1" indicating a higher degree of overfitting. Furthermore, a combination of multiple dimensions can be used to assess the level of overfitting risk.

[0048] Step S500: Determine the overfitting adjustment strategy based on the type and degree of overfitting.

[0049] The data processing method provided in this application, when the overfitting monitoring information indicates that the monitored model is overfitting, evaluates the overfitting monitoring information to obtain the overfitting type and degree. Then, based on the determined overfitting type and degree, a corresponding overfitting adjustment strategy can be determined to prepare for subsequent dynamic adjustment of the monitored model.

[0050] It is worth noting that in determining the overfitting adjustment strategy based on the overfitting type and degree, supplementary training data is first generated based on the overfitting type; then, regularization strength adjustment information, regularization method adjustment information, and training amplitude adjustment information are determined based on the overfitting degree; finally, the corresponding overfitting adjustment strategy can be determined based on the supplementary training data, regularization strength adjustment information, regularization method adjustment information, and training amplitude adjustment information, thus preparing for the subsequent dynamic adjustment of the model to be monitored.

[0051] like Figure 5 As shown, determining the overfitting adjustment strategy based on the type and degree of overfitting can include the following steps: Step S510: Generate supplementary training data according to the overfitting type; Step S520: Determine the regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information based on the degree of overfitting; Step S530: Determine the overfitting adjustment strategy based on the supplementary training data, regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information.

[0052] For steps S510 to S530, in the process of determining the overfitting adjustment strategy based on the overfitting type and degree, supplementary training data is first generated based on the overfitting type; then, regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information are determined based on the degree of overfitting; finally, the corresponding overfitting adjustment strategy can be determined based on the supplementary training data, regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information, in order to prepare for the subsequent dynamic adjustment processing of the model to be monitored.

[0053] In some specific embodiments of this application, supplementary training data is automatically generated using an intelligent data augmenter to address the detected overfitting types. A dynamic regularization controller adjusts the regularization strength and method based on the degree of overfitting. In cases of severe overfitting, a training progress regulator automatically rolls back the training steps or adjusts the learning rate. A multi-strategy optimization selector selects the optimal combination of strategies based on the overfitting type. For example, if during training the model is found to respond accurately to "loan risk" but poorly to "credit risk," it can be considered overfitting; if confidence level detection shows that the confidence difference for similar problems exceeds a threshold, it can also be considered overfitting; if business rule detection reveals rule conflicts in the handling of boundary cases, it can also be considered overfitting.

[0054] Step S600: Dynamically adjust the model to be monitored based on the overfitting adjustment strategy.

[0055] The data processing method provided in this application, after determining the overfitting adjustment strategy based on the overfitting type and degree, can dynamically adjust the monitoring model based on the overfitting adjustment strategy, so that the monitoring model can be adjusted to overcome the overfitting problem.

[0056] It is worth noting that in the process of dynamically adjusting the model to be detected based on the overfitting adjustment strategy, the model to be monitored is first optimized based on supplementary training data. Then, the model to be monitored is dynamically regularized based on the regularization intensity adjustment information and the regularization method adjustment information. Finally, the number of training steps of the model to be monitored is adjusted based on the training amplitude adjustment information. Through the above technical solution, the model to be monitored can be dynamically adjusted based on the overfitting adjustment strategy, thereby effectively overcoming the overfitting problem of the model to be monitored.

[0057] It is worth noting that, based on supplementary training data, regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information, the monitoring model can be dynamically adjusted from multiple different aspects, thereby effectively solving the overfitting problem of the monitoring model and enabling it to operate normally.

[0058] It is worth noting that, through the technical solution of this application, financial institutions can effectively control and correct overfitting problems during fine-tuning, ensuring the safety and reliability of large models in financial business applications, and providing a solid technical guarantee for the intelligent transformation of finance; the accuracy of overfitting detection is significantly improved, and multi-dimensional monitoring makes the accuracy of overfitting detection even higher; the optimization measures are highly targeted, intelligently selecting optimization strategies based on the type of overfitting, resulting in a higher effective recovery rate; and financial risks are effectively controlled, effectively preventing financial misjudgments and illegal outputs caused by overfitting. Alternatively, through the technical solution of this application, medical institutions can effectively control and correct overfitting problems during fine-tuning, ensuring the safety and reliability of large models in medical applications, and providing a solid technical guarantee for the intelligent transformation of healthcare.

[0059] like Figure 6 As shown, dynamically adjusting the model under monitoring based on the overfitting adjustment strategy can include the following steps: Step S610: Optimize the execution of the model to be monitored based on the supplementary training data; Step S620: Perform dynamic regularization control processing on the model to be monitored based on the regularization intensity adjustment information and the regularization method adjustment information; Step S630: Adjust the number of training steps of the model to be monitored based on the training amplitude adjustment information.

[0060] For steps S610 to S630, in the process of dynamically adjusting the model to be detected based on the overfitting adjustment strategy, firstly, the model to be monitored is optimized based on the supplementary training data; then, the model to be monitored is dynamically regularized based on the regularization intensity adjustment information and the regularization method adjustment information; finally, the number of training steps of the model to be monitored is adjusted based on the training amplitude adjustment information. Through the above technical solution, the model to be monitored can be dynamically adjusted based on the overfitting adjustment strategy, thereby effectively overcoming the overfitting problem of the model to be monitored.

[0061] It is worth noting that, based on supplementary training data, regularization intensity adjustment information, regularization method adjustment information, and training amplitude adjustment information, the monitoring model can be dynamically adjusted from multiple different perspectives, thereby effectively solving the overfitting problem of the monitoring model.

[0062] like Figure 7 As shown, after dynamically adjusting the model to be monitored based on the overfitting adjustment strategy, the following steps may also be included: Step S710: Obtain the overfitting monitoring and optimization report, wherein the overfitting monitoring and optimization report is obtained by dynamically adjusting the model to be monitored according to the overfitting adjustment strategy; Step S720: Transfer the overfit monitoring and optimization report to the preset overfit adjustment database.

[0063] For steps S710 to S720, after dynamically adjusting the model to be monitored based on the overfitting adjustment strategy, an overfitting monitoring optimization report can be obtained. The overfitting monitoring optimization report is obtained by dynamically adjusting the model to be monitored based on the overfitting adjustment strategy. Then, the overfitting monitoring optimization report is transferred to a preset overfitting adjustment database to record the dynamic adjustment process of the model and provide a data foundation for subsequent model operation monitoring.

[0064] It is worth noting that during the process of dynamically adjusting the model to be monitored according to the overfitting adjustment strategy, an overfitting monitoring and optimization report can be generated. Subsequently, the overfitting monitoring and optimization report can be transferred to a pre-set overfitting adjustment database to track the model adjustment process and its effects, providing a data foundation for subsequent manual decision analysis and bringing convenience to system maintenance personnel.

[0065] In addition, such as Figure 8 As shown, one embodiment of this application also provides a data processing apparatus 10, which includes: Acquisition unit 100 is used to acquire business characteristic information; Construction unit 200 is used to construct a multi-dimensional verification set based on business characteristic information and a preset hierarchical verification set construction module; The monitoring unit 300 is used to perform multi-dimensional overfitting monitoring on the model to be monitored based on the multi-dimensional validation set, and obtain overfitting monitoring information. Evaluation unit 400 is used to evaluate the overfitting monitoring information when the monitoring model is overfitted, and to obtain the overfitting type and degree. Combination unit 500 is used to determine overfitting adjustment strategies based on the type and degree of overfitting. The adjustment unit 600 is used to dynamically adjust the model to be monitored based on the overfitting adjustment strategy.

[0066] The specific implementation of the data processing device 10 is basically the same as the specific embodiment of the data processing method described above, and will not be repeated here.

[0067] In addition, such as Figure 9 As shown, one embodiment of this application also provides an electronic device 700, which includes: a memory 720, a processor 710, and a computer program stored on the memory 720 and executable on the processor 710.

[0068] The processor 710 and memory 720 can be connected via a bus or other means.

[0069] The non-transient software program and instructions required to implement the data processing method of the above embodiments are stored in the memory 720. When executed by the processor 710, the data processing method of each of the above embodiments is executed.

[0070] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0071] Furthermore, one embodiment of this application provides a computer-readable storage medium storing computer-executable instructions that are executed by a processor 710 or a controller, for example, by a processor 710 in the above-described device embodiment, causing the processor 710 to perform the data processing method in the above-described embodiment.

[0072] The above embodiments can be used in combination, and modules with the same name in different embodiments may be the same or different.

[0073] The foregoing has described specific embodiments of this application; other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than those shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily have to follow the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0074] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and computer-readable storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0075] The apparatus, device, computer-readable storage medium and method provided in the embodiments of this application are corresponding to each other. Therefore, the apparatus, device and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, device and computer storage medium will not be described again here.

[0076] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program a digital system themselves to "integrate" it onto a PLD, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used when writing program development code. The original code before compilation must also be written in a specific programming language, which is called a Hardware Description Language (HDL). There is not just one HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also understand that by simply performing some logic programming on the method flow using the aforementioned hardware description languages ​​and programming it into an integrated circuit, the hardware circuit that implements the logic method flow can be easily obtained.

[0077] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, ASICs, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0078] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0079] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, in implementing the embodiments of this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0080] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0082] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0084] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0085] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0086] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0087] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0088] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0089] The embodiments of this application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. The embodiments of this application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can reside in local and remote computer storage media, including storage devices.

[0090] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0091] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

Claims

1. A data processing method, characterized in that, The data processing method includes: Obtain business characteristic information; A multidimensional verification set is constructed based on the business characteristic information and the preset hierarchical verification set construction module; Based on the multidimensional validation set, multidimensional overfitting monitoring processing is performed on the model to be monitored to obtain overfitting monitoring information; When the overfitting monitoring information indicates that the monitored model is overfitting, the overfitting monitoring information is evaluated to obtain the overfitting type and the degree of overfitting. Determine the overfitting adjustment strategy based on the overfitting type and the degree of overfitting; The overfitting adjustment strategy is used to dynamically adjust the model to be monitored.

2. The data processing method according to claim 1, characterized in that, The construction of a multi-dimensional verification set based on the business feature information and a preset hierarchical verification set construction module includes: Based on the aforementioned business characteristic information, surface diversity information, business rule information, adversarial example information, and time stratification information are determined; Based on the surface diversity information, a surface diversity verification set is constructed using the hierarchical verification set construction module; and, based on the business rule information, a business rule verification set is constructed using the hierarchical verification set construction module; and, based on the adversarial example information, an adversarial example verification set is constructed using the hierarchical verification set construction module; and, based on the temporal hierarchical information, a temporal hierarchical verification set is constructed using the hierarchical verification set construction module. The surface diversity validation set, the business rule validation set, the adversarial example validation set, and the time-layered validation set are determined as the multidimensional validation set.

3. The data processing method according to claim 1, characterized in that, The process of performing multi-dimensional overfitting monitoring on the model to be monitored based on the multi-dimensional validation set to obtain overfitting monitoring information includes: Based on the multidimensional validation set, the model to be monitored is subjected to overfitting monitoring processing using a preset instruction consistency analyzer to obtain the first monitoring information; Based on the multidimensional validation set, the model to be monitored is subjected to overfitting monitoring processing using a preset answer diversity evaluator to obtain second monitoring information; Based on the multidimensional validation set, the model to be monitored is subjected to overfitting detection processing using a preset confidence anomaly detector to obtain third monitoring information; Based on the multidimensional validation set, the model to be monitored is subjected to overfitting monitoring processing using a preset business rule compliance checker to obtain the fourth monitoring information. The first monitoring information, the second monitoring information, the third monitoring information, and the fourth monitoring information are determined as the overfitting monitoring information.

4. The data processing method according to claim 1, characterized in that, The evaluation and processing of the overfitting monitoring information to obtain the overfitting type and degree includes: The overfitting monitoring information is segmented to obtain multiple segments of the overfitting monitoring information; Keyword recognition processing is performed on multiple overfitting segmentation monitoring information to obtain the overfitting type and the degree of overfitting.

5. The data processing method according to claim 1, characterized in that, The step of determining the overfitting adjustment strategy based on the overfitting type and the degree of overfitting includes: Generate supplementary training data based on the overfitting type; Based on the degree of overfitting, adjust the regularization intensity, regularization method, and training amplitude. The overfitting adjustment strategy is determined based on the supplementary training data, the regularization intensity adjustment information, the regularization method adjustment information, and the training amplitude adjustment information.

6. The data processing method according to claim 5, characterized in that, The dynamic adjustment of the monitored model based on the overfitting adjustment strategy includes: The monitoring model is optimized and processed based on the supplementary training data. The monitoring model is subjected to dynamic regularization control processing based on the regularization intensity adjustment information and the regularization method adjustment information; The number of training steps for the model to be monitored is adjusted based on the training amplitude adjustment information.

7. The data processing method according to claim 1, characterized in that, After dynamically adjusting the model to be monitored based on the overfitting adjustment strategy, the method further includes: Obtain an overfitting monitoring and optimization report, wherein the overfitting monitoring and optimization report is obtained by dynamically adjusting the model to be monitored according to the overfitting adjustment strategy; The overfit monitoring and optimization report is transferred to a preset overfit adjustment database.

8. A data processing apparatus, characterized in that, The data processing device includes: The acquisition unit is used to acquire business characteristic information; The construction unit is used to construct a multidimensional verification set based on the business feature information and the preset hierarchical verification set construction module; The monitoring unit is used to perform multi-dimensional overfitting monitoring processing on the model to be monitored based on the multi-dimensional validation set, and obtain overfitting monitoring information. An evaluation unit is used to evaluate the overfitting monitoring information when the overfitting monitoring information indicates that the monitored model is overfitting, and to obtain the overfitting type and the degree of overfitting. A combination unit is used to determine an overfitting adjustment strategy based on the overfitting type and the degree of overfitting. The adjustment unit is used to dynamically adjust the model to be monitored based on the overfitting adjustment strategy.

9. An electronic device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the data processing method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are used to execute the data processing method according to any one of claims 1 to 7.

Images