Adaptive detection rule optimization method and device

Abstract

The embodiment of the invention provides an adaptive detection rule optimization method and device, and the method comprises the steps: receiving an adjustment and optimization instruction and a current detection rule, and receiving adjustment and optimization data and a data type corresponding to each piece of adjustment and optimization data; inputting the tuning data, the data category and the current detection rule into the tuning model to obtain an updated detection rule, obtaining a standard test set, processing the standard test set through the updated detection rule to obtain a first test result, and processing the standard test set through the detection rule to obtain a second test result; displaying the first test result and the second test result on a display screen, receiving an adjusting and optimizing selection instruction, adjusting model parameters of the adjusting and optimizing model based on the adjusting and optimizing selection instruction, and repeatedly executing the above steps through the adjusting and optimizing model with the adjusted model parameters until it is indicated that updating of the detection rule is stopped. According to the method, the defects that the adjustment and optimization process of the detection rule is tedious and the expansibility is poor are effectively overcome, and the dynamic adjustment and optimization convenience degree and the dynamic requirement of the detection rule are remarkably improved.

Description

Technical Field

[0001] This application relates to the field of data processing, specifically to an adaptive detection rule optimization method and apparatus. Background Technology

[0002] With the development of artificial intelligence, the amount of content in Software-as-a-Service (SaaS) is growing explosively, and at the same time, the review scenarios are becoming increasingly complex.

[0003] In existing technologies, SaaS sets up a series of standardized detection rules that can be used for scenarios such as content review, compliance checks, and asset classification. These detection rules can provide basic services to different users in a multi-tenant environment. However, each user, and even different users within each user group, generally have unique and subtle differences in their needs for the detection rules, and the detection rules need to be updated in real time.

[0004] For differentiated needs and real-time updates, users need to optimize existing detection rules. Currently, users need to invite professional rule engineers to participate in the optimization, but the optimization process is cumbersome, has a long response cycle, and the detection rules have poor scalability, making it difficult to meet the dynamic optimization needs of a large number of users on a large scale. Summary of the Invention

[0005] To address the problems in the prior art, this application provides an adaptive detection rule optimization method and apparatus, which can effectively solve the shortcomings of traditional technologies such as cumbersome detection rule tuning process, high cost and poor scalability, and significantly improve the convenience of dynamic tuning of detection rules and meet dynamic needs.

[0006] To solve at least one of the above problems, this application provides the following technical solution:

[0007] Firstly, this application provides an adaptive detection rule optimization method, including:

[0008] Receive tuning instructions, determine rule groups and corresponding current detection rules based on tuning instructions, and receive tuning data and corresponding data categories for each tuning data. Tuning data includes text data and image data, with a one-to-one correspondence between text data and image data. Data categories include positive example category, negative example category, and suspected example category.

[0009] Input the optimization data, data category and current detection rules into the optimization model to obtain the updated detection rules, obtain the standard test set, process the standard test set through the updated detection rules to obtain the first test result, and process the standard test set through the detection rules to obtain the second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time.

[0010] The first test result and the updated detection rule, the second test result and the detection rule are displayed on the screen. The tuning selection instruction is received, and the model parameters of the tuning model are adjusted based on the tuning selection instruction. The tuning model with the adjusted model parameters is used to repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule, until the tuning selection instruction indicates to stop updating the detection rule, and the updated detection rule is saved to the current rule group.

[0011] Furthermore, it also includes: receiving image data or text data, and sending annotation prompt information when receiving image data or text data, receiving prompt feedback information corresponding to the annotation prompt message, the annotation prompt information being used to prompt the user to upload text data corresponding to the image data, or to upload image data corresponding to the text data;

[0012] Image data and the corresponding text data are defined as a set of tuning arrays. The tuning data includes one or more tuning arrays.

[0013] Furthermore, it also includes: preprocessing the optimization data and extracting the features of the preprocessed optimization data to obtain optimization data features, which include text data features and image data features;

[0014] Text data features are input into the text branch of the optimization model for processing to obtain text processing data. Image data features are input into the image branch of the optimization model for processing to obtain image processing data. The rule data features corresponding to the current detection rule stored in the optimization model are retrieved.

[0015] By dynamically adjusting the weights of the features of text processing data, image processing data, and rule data through a gating attention mechanism, and then performing a weighted average of the features of text processing data, image processing data, and rule data, the updated detection rules are obtained.

[0016] Furthermore, it also includes: determining the optimization result of the current round based on the optimization selection instruction, wherein the optimization result is either the first test result or the second test result;

[0017] If the optimization result is the first test result, the first test result is determined as the reward signal for the current round, and the second test result is determined as the penalty signal for the current round.

[0018] If the optimization result is the second test result, the second test result will be determined as the reward signal for the current round, and the first test result will be determined as the penalty signal for the current round.

[0019] The learning rate of the tuning model is increased based on reward signals and decreased based on penalty signals in order to adjust the model parameters.

[0020] Furthermore, tuning instructions include adopting update instructions, continuing optimization instructions, and reverting to previous versions; after receiving tuning selection instructions, it also includes:

[0021] When the tuning selection instruction is to adopt the update instruction, stop updating the detection rules and the model parameters of the tuning model;

[0022] When the tuning selection instruction is to continue optimization, the steps of adjusting the model parameters of the tuning model based on the tuning selection instruction are executed.

[0023] When the tuning selection instruction is a rollback instruction, the detection rule corresponding to the previous iteration is retrieved, and the detection rule corresponding to the previous iteration is determined as the updated detection rule.

[0024] Furthermore, it also includes: receiving tuning instructions through an interactive interface on the display screen;

[0025] After receiving the tuning instructions, it also includes:

[0026] The basic information of the current detection rule and the historical performance data of the current detection rule are displayed on the terminal device's screen. The basic information includes the rule name and rule description of the detection rule.

[0027] Display the historical data to be detected within the preset time period before the current time, processed by the detection rules, and the corresponding historical detection results on the display screen.

[0028] Secondly, this application provides an adaptive detection rule optimization device, comprising:

[0029] The receiving module is used to receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between text data and image data. The data categories include positive example category, negative example category and suspected example category.

[0030] The testing module is used to input the tuning data, data category and current detection rules into the tuning model to obtain the updated detection rules, obtain the standard test set, process the standard test set through the updated detection rules to obtain the first test result, and process the standard test set through the detection rules to obtain the second test result. The standard test set includes historical sample data and tuning sample data received within a preset time period before the current time.

[0031] The tuning module is used to display the first test result and the updated detection rules, the second test result and the detection rules on the display screen, receive tuning selection instructions, adjust the model parameters of the tuning model based on the tuning selection instructions, and repeatedly execute the steps of inputting tuning data, data category and current detection rules into the tuning model to obtain updated detection rules through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rules, and save the updated detection rules to the current rule group.

[0032] Thirdly, this application provides 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 program to implement the steps of the adaptive detection rule optimization method.

[0033] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the adaptive detection rule optimization method described above.

[0034] Fifthly, this application provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the adaptive detection rule optimization method.

[0035] As can be seen from the above technical solution, this application provides an adaptive detection rule optimization method and apparatus. It innovatively receives tuning instructions, determines rule groups and corresponding current detection rules based on these instructions, and receives tuning data and corresponding data categories. The tuning data includes text data and image data, with a one-to-one correspondence between them. The tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules. Simultaneously, a standard test set is acquired. The updated detection rules are used to process the standard test set to obtain a first detection result, and the previous detection rules are used to process the standard test set to obtain a second test result. The first test result and the updated detection rules, as well as the second test result and the previous detection rules, are displayed on a screen. A tuning selection instruction is received, and the model parameters of the tuning model are adjusted according to the instruction. The tuning process is repeated using the tuning model corresponding to the adjusted model parameters until the tuning selection instruction indicates that updating the detection rules should be stopped. The updated detection rules are then saved to the current rule group. This method effectively addresses the shortcomings of cumbersome, costly, and poorly scalable optimization processes in detection rule tuning, significantly improving the convenience and responsiveness of dynamic rule tuning. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating the adaptive detection rule optimization method in the embodiments of this application;

[0038] Figure 2 This is a structural diagram of the adaptive detection rule optimization device in the embodiments of this application;

[0039] Figure 3 This is a schematic diagram of the structure of the electronic device in the embodiments of this application.

[0040] Figure label:

[0041] Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver storage unit 9144, antenna 9111, speaker 9131, microphone 9132. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0044] In existing technologies, general and unchanging detection rules are difficult to accurately meet the personalized needs of specific users, resulting in unsatisfactory accuracy and recall of detection results. If it is necessary to manually adjust the detection rules for each user, professional rule engineers are required, and the adjustment process is cumbersome, has a long response cycle, and poor scalability, making it impossible to meet the dynamic optimization needs of a large number of users on a large scale.

[0045] This application provides an adaptive rule-based detection optimization method. This method draws on the ideas of human-in-the-loop and continuous learning to construct a self-optimizing closed-loop method for detection rules. The core idea is to combine the detection results of the detection rules with the user's professional judgment. The corresponding embodiment of this method can provide an intuitive interactive interface to guide the user to "teach" the detection sample data (i.e., provide positive examples, negative examples, and uncertain examples) and explain the basis for judgment. It automatically captures user feedback and uses it as the core input to iterate and generate a new, optimized detection rule that better meets the user's expectations.

[0046] To effectively address the shortcomings of traditional technologies, such as cumbersome and costly optimization processes and poor scalability, and to significantly improve the convenience and responsiveness of dynamic rule optimization, this application provides an embodiment of an adaptive detection rule optimization method. (See attached example.) Figure 1 The adaptive detection rule optimization method specifically includes the following:

[0047] Step S101: Receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data.

[0048] The optimization data includes text data and image data, with a one-to-one correspondence between the text data and the image data. The data categories include positive example category, negative example category, and suspected example category.

[0049] Optionally, this embodiment receives an optimization instruction, and can initiate the optimization process of the detection rules according to the optimization instruction. Based on the optimization instruction, the rule group to which the detection rule to be optimized belongs is determined, and the detection rules corresponding to the current rule group are determined. The rule group is used to represent different detection rules corresponding to different needs. For example, the detection rules included in the content review group are different from the detection rules included in the compliance inspection group, and can be distinguished by the rule group.

[0050] Furthermore, all current detection rules can be optimized from one or a group of default detection rules, and can be saved to different rule groups according to the different functions of the detection rules. The rule groups can be predefined or dynamically generated according to user needs.

[0051] In addition, the system receives optimization data and the corresponding data categories for each optimization data. The optimization data includes text data and image data, with a one-to-one correspondence between the text data and the image data, which enables the text data and image data to be correctly associated and processed during the optimization process. The data categories for optimization data include positive example categories, negative example categories, and suspected example categories.

[0052] Among them, optimization data can be obtained in various ways, such as by users uploading from the interactive interface, automatic collection, or by obtaining data from third-party data sources. Automatic collection methods include automatically adding the data to the content pool from the Digital Asset Management (DAM) system whose test results have been corrected by users.

[0053] Among them, the positive example category is used to represent the optimization data that meets the expectations of the detection rules, the negative example category is used to represent the optimization data that does not meet the expectations of the detection rules, and the suspected example category includes optimization data that is difficult to classify directly and requires further analysis.

[0054] Furthermore, automated data labeling tools can be used to perform preliminary classification on newly collected optimization data using machine learning models to obtain preliminary classification results. These preliminary classification results can then be displayed on a screen, and confirmation and adjustment information can be received to obtain the adjusted classification results.

[0055] This embodiment enables the initiation of the tuning process based on tuning instructions and the receipt of tuning data from the instruction set, thereby improving the quality of tuning.

[0056] Step S102: Input the optimization data, data category and current detection rules into the optimization model to obtain the updated detection rules, obtain the standard test set, process the standard test set through the updated detection rules to obtain the first test result, and process the standard test set through the detection rules to obtain the second test result.

[0057] The standard test set includes historical sample data and optimized sample data received within a preset time period prior to the current time.

[0058] Optionally, in this embodiment, the optimization data, data categories, and detection rules are input into the optimization model, so that the optimization model optimizes the current detection rules based on the above data to obtain updated detection rules. The updated detection rules may include adjusting keyword weights, adding image texture features, etc., so that the detection rules can better learn the associations reflected in the positive example categories and avoid associations in the negative example categories.

[0059] In addition, a standard test set is obtained, which includes historical sample data and optimized sample data received within a preset time period before the current time. The standard test set is processed by the updated detection rules to obtain the first test result, and the standard test set is processed by the unupdated detection rules to obtain the second test result.

[0060] The update performance of the updated detection rule can be determined by the first test result and the second test result. Furthermore, the standard test set can be divided into multiple subsets by cross-validation technology, and the first test result and the second test result can be determined multiple times to obtain a more reliable performance evaluation.

[0061] This embodiment implements the testing of updated detection rules with a standard test set to obtain the first test result, which can quickly provide feedback on the performance after the detection rules are updated and improve the user experience.

[0062] Step S103: Display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive the tuning selection instruction, adjust the model parameters of the tuning model based on the tuning selection instruction, and repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

[0063] Optionally, in this embodiment, the first test result and the updated detection rule, and the second test result and the detection rule before the update are displayed on the display screen. The first test result and the updated detection rule can be displayed as one set of data, and the second test result and the detection rule before the update can be displayed as another set of data on the display screen, so that the user can compare and select through the display screen.

[0064] In addition, the system receives tuning selection instructions, meaning that users can select the required content based on the first test result and the updated detection rules, the second test result and the detection rules before the update, displayed on the screen, and send tuning selection instructions based on the selected content. After receiving the tuning selection instructions, the system adjusts the model parameters of the tuning model according to the tuning selection instructions.

[0065] Among them, adjusting and optimizing the model parameters can be done through hyperparameter optimization in machine learning, such as through network search, random search, or Bayesian optimization.

[0066] By repeatedly executing the steps of inputting optimization data, data categories, and current detection rules into the optimization model using the adjusted model parameters, the process continues until the optimization selection command indicates that the update of the detection rules should be stopped. This allows for multiple iterations and updates of the detection rules until they meet the user's expectations. The optimization selection command allows the user to adjust the direction of the model parameters.

[0067] In addition, the updated detection rules are saved to the current rule group to obtain the expected detection rules and meet personalized needs.

[0068] Furthermore, the data categories and data types of the optimization data can be adjusted during the optimization process. For example, the suspected case category can be changed to the positive case category or the negative case category. Optimization data can also be added or reduced.

[0069] This embodiment enables the adjustment of the optimized model by adjusting model parameters, thereby updating the detection rules and improving the applicability, personalization, and user satisfaction of the updated detection rules.

[0070] This embodiment achieves efficient optimization of detection rules, improves the accuracy of updated detection rules and the adaptability of the tuning model. Through user-uploaded tuning instructions, tuning selection instructions and tuning data, detection rules can be adjusted simply and directly, improving the flexibility of detection rule updates, reducing the cost of updating detection rules and enhancing the user experience.

[0071] In some embodiments, receiving tuning data and the data category corresponding to each tuning data includes:

[0072] Receive image data or text data, and send annotation prompt information when receiving image data or text data, and receive prompt feedback information corresponding to the annotation prompt message. The annotation prompt information is used to prompt the user to upload text data corresponding to the image data, or upload image data corresponding to the text data.

[0073] Image data and the corresponding text data are defined as a set of tuning arrays. The tuning data includes one or more tuning arrays.

[0074] Optionally, this embodiment receives image data or text data. When image data is received, a notification message is sent prompting the upload of the corresponding text data. Conversely, when text data is received, a notification message is also sent prompting the upload of the corresponding image data.

[0075] The sending and receiving of image and text data with annotation prompts can be done through a visual interface on the display screen. The annotation prompts are used to guide users to upload text or image data corresponding to the image or text data, so as to keep the optimization data guaranteed and consistent. The annotation prompts can improve the quality and relevance of the optimization data and improve the accuracy of the updated detection rules.

[0076] Furthermore, natural language processing and computer vision can be used to assist users in annotating the optimized data. For example, pre-trained natural language processing models can automatically generate summaries or keywords for text data, or computer vision can automatically extract feature descriptions from image data to help users quickly complete the annotation of optimized data and reduce their workload.

[0077] In addition, the image data and the corresponding text data are defined as a set of tuning arrays. The tuning data may include one or more sets of tuning arrays, which can be determined according to user needs and configuration.

[0078] Furthermore, preprocessing operations are performed on the received image and text data. These preprocessing operations include, but are not limited to, resizing the image data and segmenting and standardizing the text data to ensure consistency and processability of the optimized data.

[0079] This embodiment achieves efficient reception and processing of optimization data. It can ensure the integrity and consistency of optimization data through annotation prompt messages, thereby improving the processing efficiency of optimization data and enhancing the user experience.

[0080] In some embodiments, tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules, including:

[0081] The optimization data is preprocessed, and the features of the preprocessed optimization data are extracted to obtain optimization data features, which include text data features and image data features.

[0082] Text data features are input into the text branch of the optimization model for processing to obtain text processing data. Image data features are input into the image branch of the optimization model for processing to obtain image processing data. The rule data features corresponding to the current detection rule stored in the optimization model are retrieved.

[0083] By dynamically adjusting the weights of the features of text processing data, image processing data, and rule data through a gating attention mechanism, and then performing a weighted average of the features of text processing data, image processing data, and rule data, the updated detection rules are obtained.

[0084] Optionally, this embodiment preprocesses the tuning data. The preprocessing of text data includes, but is not limited to, word segmentation, stop word removal, stemming, and word form restoration. The preprocessing of image data includes, but is not limited to, size standardization, color space conversion, noise removal, and feature point detection. The preprocessed tuning data can transform the original tuning data into a form that is easier for the tuning model to process and understand.

[0085] In addition, features of the processed optimization data are extracted to obtain optimization data features, which include text data features and image data features. Text data features include, but are not limited to, word frequency, TF-IDF value, and word embedding vector. Image data features include, but are not limited to, color histogram, texture features, and high-level features extracted by deep learning models.

[0086] In addition, the text data features are input into the text branch of the tuning model for processing to obtain text processing data. The model involved in the text branch can be a recurrent neural network (RNN) model or a transformer deep learning model.

[0087] Simultaneously, the image data features are input into the image branch of the tuning model for processing to obtain image processing data. The model involved in the image branch can be a convolutional neural network (CNN) model.

[0088] At the same time, retrieve the rule data features corresponding to the current detection rule stored in the current detection group in the optimization model.

[0089] Furthermore, the weights corresponding to the features of text processing data, image processing data, and rule data are dynamically adjusted through a gating attention mechanism. The gating attention mechanism is a mechanism that can automatically adjust the weights according to the importance of the input data, which can make the tuning model pay more attention to information that is more helpful in tuning the current detection rule.

[0090] Furthermore, by weighted averaging of features from text processing data, image processing data, and rule-based data, updated detection rules are obtained. This allows for the integration of data features from all data types, generating a new, more accurate, and comprehensive detection rule.

[0091] Furthermore, the gating attention mechanism can dynamically adjust weights based on the contextual information of the optimized data. For example, when image data has a greater impact on the detection results, higher weights can be assigned to image processing data.

[0092] This embodiment improves the accuracy and robustness of detection rules through feature extraction and multimodal fusion, enhances the flexibility and focus of the tuning model through the gating attention mechanism, improves the accuracy of updated detection rules, optimizes computational efficiency, and meets users' needs for the accuracy and flexibility of detection rules.

[0093] In some embodiments, adjusting the model parameters of the tuning model based on tuning selection instructions includes:

[0094] The tuning result for the current round is determined based on the tuning selection instruction. The tuning result is either the first test result or the second test result.

[0095] If the optimization result is the first test result, the first test result is determined as the reward signal for the current round, and the second test result is determined as the penalty signal for the current round.

[0096] If the optimization result is the second test result, the second test result will be determined as the reward signal for the current round, and the first test result will be determined as the penalty signal for the current round.

[0097] The learning rate of the tuning model is increased based on reward signals and decreased based on penalty signals in order to adjust the model parameters.

[0098] Optionally, this embodiment receives a tuning selection instruction, which is used to indicate the tuning result of the current round. The tuning result can be a first test result or a second test result, which correspond to the results of processing the standard test set using the updated detection rules and the original detection rules, respectively.

[0099] The first test results and the second test results may include, but are not limited to, the positive case expectation rate, the negative case expectation rate, and the suspected case expectation rate.

[0100] Furthermore, based on the optimization selection instruction, the optimization result for the current round is determined. If the optimization result is the first test result (i.e., obtained through the updated detection rules), the first test result is used as the reward signal for the current round, and the second test result is used as the penalty signal for the current round. Conversely, if the optimization result is the second test result (i.e., obtained through the original detection rules), the second test result is used as the reward signal, and the first test result is used as the penalty signal.

[0101] Furthermore, the learning rate of the tuning model is increased based on reward signals to accelerate the learning process, while the learning rate is decreased based on penalty signals to slow down the learning speed. This allows for dynamic adjustment of the learning rate based on the tuning results, enabling flexible adjustment of the training intensity of the tuning model.

[0102] Furthermore, the tuning process can be viewed as a reinforcement learning task, where the parameter adjustment of the model is similar to the action selection of an agent, and reward and penalty signals can be used as reward values ​​in reinforcement learning to guide the adjustment of model parameters.

[0103] Furthermore, in the process of adjusting model parameters, in addition to considering a single reward signal or penalty signal, a multi-objective optimization mechanism can be introduced to simultaneously consider multiple objectives such as the accuracy, robustness, recall, and computational efficiency of the detection rules.

[0104] This embodiment enables the adjustment of model parameters based on tuning selection instructions. It can dynamically adjust the learning rate according to the tuning results of the current round, quickly respond to the feedback of tuning results, improve tuning efficiency, enhance the adaptability of the tuned model, and optimize model performance. In other words, it can be gradually optimized in multiple iterations to achieve better performance and improve the overall operating efficiency and accuracy of the tuned model.

[0105] In some embodiments, the tuning instructions include an adoption update instruction, a continue optimization instruction, and a rollback version instruction; after receiving a tuning selection instruction, it further includes:

[0106] When the tuning selection instruction is to adopt the update instruction, stop updating the detection rules and the model parameters of the tuning model;

[0107] When the tuning selection instruction is to continue optimization, the steps of adjusting the model parameters of the tuning model based on the tuning selection instruction are executed.

[0108] When the tuning selection instruction is a rollback instruction, the detection rule corresponding to the previous iteration is retrieved, and the detection rule corresponding to the previous iteration is determined as the updated detection rule.

[0109] Optionally, the optimization selection instructions in this embodiment include, but are not limited to, adopt update instructions, continue optimization instructions, and rollback instructions. Different processing logic can be adopted according to different optimization selection instructions. Different optimization selection instructions are used to finely control the detection rules according to the test results and business requirements.

[0110] When an update acceptance instruction is received, the updating of detection rules and model parameters of the tuning model are stopped. In other words, accepting the update instruction means that the user is satisfied with the current detection rules. At the same time, the updated detection rules are saved and deployed to the production environment.

[0111] Furthermore, a version control system can be used to track each iteration and change of the detection rules. This system can help users go back to any historical version and provide detailed logs of changes to the detection rules.

[0112] When a continued optimization instruction is received, the steps of adjusting the model parameters of the tuned model based on the optimization selection instruction are executed. This includes determining the optimization result of the current round, adjusting the learning rate according to the optimization result, and continuing to optimize the detection rules and model parameters. Hyperparameters of the tuned model, such as the learning rate, can also be adjusted through reward and penalty signals.

[0113] Furthermore, through meta-learning, the tuning model can automatically adjust its optimization strategy based on previous tuning experience. Meta-learning can help the tuning model adapt to new tuning data and environments more quickly, thereby improving the efficiency of the tuning process.

[0114] When a rollback command is received, the detection rules corresponding to the previous iteration are retrieved and identified as the updated detection rules. The rollback command can be used to undo the optimization operations of the current iteration and restore the detection rules of the previous version. In other words, it can quickly restore to the previous stable version if the updated detection rules are not performing well.

[0115] Furthermore, a user-friendly interface can be provided to allow users to browse different versions of the detection rules and select the version they want to revert to. In addition, a version comparison tool can be provided to help users understand the differences between different versions of the detection rules.

[0116] This embodiment enables the control of the detection rule change process through tuning commands, improving the controllability of the detection rules and user satisfaction. It can also quickly adjust the detection rules based on user feedback and business needs, enhancing flexibility and adaptability, thereby better meeting users' personalized needs while maintaining the stability and reliability of the detection rules, thus improving user experience and satisfaction.

[0117] In some embodiments, receiving tuning instructions includes:

[0118] Receive tuning instructions through the interactive interface on the display screen;

[0119] After receiving the tuning instructions, it also includes:

[0120] The basic information of the current detection rule and the historical performance data of the current detection rule are displayed on the terminal device's screen. The basic information includes the rule name and rule description of the detection rule.

[0121] Display the historical data to be detected within the preset time period before the current time, processed by the detection rules, and the corresponding historical detection results on the display screen.

[0122] Optionally, in this embodiment, receiving tuning instructions is a step in realizing user participation and feedback loop, and the user's tuning instructions can be received through the interactive interface on the display screen.

[0123] The interactive interface can be a graphical user interface (GUI) that allows users to trigger tuning commands by clicking buttons, sliders, or other controls. Furthermore, the interactive interface can also provide text input boxes and image upload boxes to upload tuning data. It can also display data categories on the interactive interface so that users can directly upload tuning data to the corresponding data category. For example, users can drag and drop images and text of the positive example category or click to upload them to the position of the positive example category displayed on the interactive interface.

[0124] Furthermore, a text input box can be provided on the interactive interface, allowing users to input more detailed instructions or feedback, such as feedback on the first or second test results, which can be used to further adjust and optimize the model based on the received text feedback.

[0125] In addition, after receiving the tuning instructions, the basic information and historical performance data of the current detection rule are displayed on the terminal device's screen. The basic information includes the rule name and rule description of the detection rule to help users understand the function and purpose of the current detection rule, while the historical performance data helps users understand the specific capabilities of the current detection rule.

[0126] In addition, the display screen can also show the historical data to be detected and the historical detection results processed by the detection rules within a preset time period before the current time. The historical data to be detected and the historical detection results may include, but are not limited to, the performance indicators such as the number of samples processed by the detection rules, accuracy, and recall, as well as the specific sample processing results.

[0127] Furthermore, data visualization techniques, such as charts, graphs, and dashboards, can be used to visually demonstrate the performance and effectiveness of detection rules. For example, line graphs can be used to show how the performance of detection rules changes over time, or heatmaps can be used to show how different detection rules perform on different datasets.

[0128] This embodiment achieves better satisfaction of users' personalized needs by increasing user participation, information transparency, and decision-making efficiency, while maintaining stability and reliability, thus improving overall performance and user experience.

[0129] To effectively address the shortcomings of traditional technologies, such as cumbersome detection rule tuning processes, high costs, and poor scalability, and to significantly improve the convenience and responsiveness of dynamic detection rule tuning, this application provides an embodiment of an adaptive detection rule optimization device for implementing all or part of the adaptive detection rule optimization. See [link to embodiment]. Figure 2 The adaptive detection rule optimization device specifically includes the following components:

[0130] The receiving module 10 is used to receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between text data and image data. The data categories include positive example category, negative example category and suspected example category.

[0131] The testing module 20 is used to input the optimization data, data category and current detection rules into the optimization model to obtain the updated detection rules, obtain the standard test set, process the standard test set through the updated detection rules to obtain the first test result, and process the standard test set through the detection rules to obtain the second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time.

[0132] The tuning module 30 is used to display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive tuning selection instructions, adjust the model parameters of the tuning model based on the tuning selection instructions, and repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

[0133] As described above, the adaptive detection rule optimization device provided in this application embodiment can innovatively receive tuning instructions, determine rule groups and corresponding current detection rules based on the tuning instructions, and receive tuning data and corresponding data categories. The tuning data includes text data and image data, with a one-to-one correspondence between them. The tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules. Simultaneously, a standard test set is acquired, and the updated detection rules are used to process the standard test set to obtain a first detection result. The updated detection rules are then used to process the standard test set to obtain a second test result. The first test result and the updated detection rules, as well as the second test result and the updated detection rules, are displayed on a screen. The device also receives tuning selection instructions, adjusts the model parameters of the tuning model according to the instructions, and repeatedly executes the tuning process using the tuning model corresponding to the adjusted model parameters until the tuning selection instructions indicate that updating the detection rules should be stopped. Finally, the updated detection rules are saved to the current rule group. This method effectively addresses the shortcomings of cumbersome, costly, and poorly scalable optimization processes in detection rule tuning, significantly improving the convenience and responsiveness of dynamic rule tuning.

[0134] From a hardware perspective, in order to effectively address the shortcomings of traditional technologies, such as cumbersome detection rule tuning processes, high costs, and poor scalability, and to significantly improve the convenience and responsiveness of dynamic detection rule tuning, this application provides an embodiment of an electronic device for implementing all or part of the adaptive detection rule optimization method. The electronic device specifically includes the following components:

[0135] The system comprises a processor, memory, a communications interface, and a bus; wherein the processor, memory, and communications interface communicate with each other via the bus; the communications interface is used to realize information transmission between the adaptive detection rule optimization device and core business systems, user terminals, and related databases and other related devices; the logic controller can be a desktop computer, tablet computer, or mobile terminal, etc., and this embodiment is not limited to these. In this embodiment, the logic controller can be implemented with reference to the embodiments of the adaptive detection rule optimization method and the adaptive detection rule optimization device in the embodiments, the contents of which are incorporated herein, and repeated details will not be described again.

[0136] It is understood that the user terminal may include smartphones, tablet computers, network set-top boxes, portable computers, desktop computers, personal digital assistants (PDAs), in-vehicle devices, smart wearable devices, etc. Among these, the smart wearable devices may include smart glasses, smartwatches, smart bracelets, etc.

[0137] In practical applications, the adaptive detection rule optimization method can be partially executed on the electronic device side as described above, or all operations can be completed on the client device. The choice can be made based on the processing power of the client device and the limitations of the user's usage scenario. This application does not impose any limitations on this. If all operations are completed on the client device, the client device may further include a processor.

[0138] The aforementioned client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission. The server may include a server on the task scheduling center side; in other implementation scenarios, it may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, a server cluster consisting of multiple servers, or a distributed server structure.

[0139] Figure 3 This is a schematic block diagram illustrating the system configuration of the electronic device 9600 according to an embodiment of this application. Figure 3 As shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It is worth noting that... Figure 3 This is an example; other types of structures can also be used to supplement or replace this structure to achieve telecommunications functions or other functions.

[0140] In one embodiment, the adaptive detection rule optimization method function can be integrated into the central processing unit 9100. The central processing unit 9100 can be configured to perform the following control:

[0141] Step S101: Receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between text data and image data. The data categories include positive example category, negative example category and suspected example category.

[0142] Step S102: Input the optimization data, data category and current detection rule into the optimization model to obtain the updated detection rule, obtain the standard test set, process the standard test set through the updated detection rule to obtain the first test result, and process the standard test set through the detection rule to obtain the second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time.

[0143] Step S103: Display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive the tuning selection instruction, adjust the model parameters of the tuning model based on the tuning selection instruction, and repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

[0144] As described above, the electronic device provided in this application embodiment innovatively receives tuning instructions, determines rule groups and corresponding current detection rules based on the tuning instructions, and receives tuning data and corresponding data categories. The tuning data includes text data and image data, with a one-to-one correspondence between them. The tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules. Simultaneously, a standard test set is acquired, and the updated detection rules are used to process the standard test set to obtain a first detection result. The updated detection rules are then used to process the standard test set to obtain a second test result. The first test result and the updated detection rules, as well as the second test result and the updated detection rules, are displayed on a screen. A tuning selection instruction is received, and the model parameters of the tuning model are adjusted according to the tuning selection instruction. The tuning process is repeated using the tuning model corresponding to the adjusted model parameters until the tuning selection instruction indicates that the update of the detection rules should be stopped. The updated detection rules are then saved to the current rule group. This method effectively addresses the shortcomings of cumbersome, costly, and poorly scalable optimization processes in detection rule tuning, significantly improving the convenience and responsiveness of dynamic rule tuning.

[0145] In another embodiment, the adaptive detection rule optimization device can be configured separately from the central processing unit 9100. For example, the adaptive detection rule optimization device can be configured as a chip connected to the central processing unit 9100, and the adaptive detection rule optimization method function can be implemented through the control of the central processing unit.

[0146] like Figure 3 As shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily need to include these components. Figure 3 All components shown; in addition, the electronic device 9600 may also include Figure 3 For components not shown, please refer to existing technologies.

[0147] like Figure 3 As shown, the central processing unit 9100, sometimes also referred to as a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives inputs and controls the operation of various components of the electronic device 9600.

[0148] The memory 9140 may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store a program for executing that information. The central processing unit 9100 may execute the program stored in the memory 9140 to perform information storage or processing, etc.

[0149] Input unit 9120 provides input to central processing unit 9100. Input unit 9120 may be, for example, a keypad or touch input device. Power supply 9170 provides power to electronic device 9600. Display 9160 displays images and text. Display may be, for example, an LCD display, but is not limited thereto.

[0150] The memory 9140 can be a solid-state memory, such as a read-only memory (ROM), random access memory (RAM), a SIM card, etc. It can also be a memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes referred to as EPROMs. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142 for storing application programs and function programs or processes for executing the operation of the electronic device 9600 via the central processing unit 9100.

[0151] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the electronic device's communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0152] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processing unit 9100 to provide input signals and receive output signals, which is the same as in a conventional mobile communication terminal.

[0153] Based on different communication technologies, multiple communication modules 9110 can be configured in the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide audio output via the speaker 9131 and receive audio input from the microphone 9132, thereby realizing typical telecommunications functions. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. Additionally, the audio processor 9130 is coupled to a central processing unit 9100, enabling on-device recording via the microphone 9132 and on-device playback of stored audio via the speaker 9131.

[0154] Embodiments of this application also provide a computer-readable storage medium capable of implementing all steps of the adaptive detection rule optimization method with a server or client execution subject in the above embodiments. The computer-readable storage medium stores a computer program that, when executed by a processor, implements all steps of the adaptive detection rule optimization method with a server or client execution subject in the above embodiments. For example, when the processor executes the computer program, it implements the following steps:

[0155] Step S101: Receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between text data and image data. The data categories include positive example category, negative example category and suspected example category.

[0156] Step S102: Input the optimization data, data category and current detection rule into the optimization model to obtain the updated detection rule, obtain the standard test set, process the standard test set through the updated detection rule to obtain the first test result, and process the standard test set through the detection rule to obtain the second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time.

[0157] Step S103: Display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive the tuning selection instruction, adjust the model parameters of the tuning model based on the tuning selection instruction, and repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

[0158] As described above, the computer-readable storage medium provided in this application embodiment innovatively receives tuning instructions, determines rule groups and corresponding current detection rules based on the tuning instructions, and receives tuning data and corresponding data categories. The tuning data includes text data and image data, with a one-to-one correspondence between them. The tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules. Simultaneously, a standard test set is acquired, and the updated detection rules are used to process the standard test set to obtain a first detection result. The updated detection rules are then used to process the standard test set to obtain a second test result. The first test result and the updated detection rules, as well as the second test result and the updated detection rules, are displayed on a screen. A tuning selection instruction is received, and the model parameters of the tuning model are adjusted according to the tuning selection instruction. The tuning process is repeated using the tuning model corresponding to the adjusted model parameters until the tuning selection instruction indicates that the update of the detection rules should be stopped. The updated detection rules are then saved to the current rule group. This method effectively addresses the shortcomings of cumbersome, costly, and poorly scalable optimization processes in detection rule tuning, significantly improving the convenience and responsiveness of dynamic rule tuning.

[0159] Embodiments of this application also provide a computer program product capable of implementing all steps of the adaptive detection rule optimization method with the execution subject being a server or client in the above embodiments. When this computer program / instruction is executed by a processor, it implements the steps of the adaptive detection rule optimization method. For example, the computer program / instruction implements the following steps:

[0160] Step S101: Receive tuning instructions, determine the rule group and the current detection rule corresponding to the rule group based on the tuning instructions, and receive tuning data and the data category corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between text data and image data. The data categories include positive example category, negative example category and suspected example category.

[0161] Step S102: Input the optimization data, data category and current detection rule into the optimization model to obtain the updated detection rule, obtain the standard test set, process the standard test set through the updated detection rule to obtain the first test result, and process the standard test set through the detection rule to obtain the second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time.

[0162] Step S103: Display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive the tuning selection instruction, adjust the model parameters of the tuning model based on the tuning selection instruction, and repeatedly execute the steps of inputting tuning data, data category and current detection rule into the tuning model to obtain the updated detection rule through the tuning model with adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

[0163] As described above, the computer program product provided in this application innovatively receives tuning instructions, determines rule groups and corresponding current detection rules based on the tuning instructions, and receives tuning data and corresponding data categories. The tuning data includes text data and image data, with a one-to-one correspondence between them. The tuning data, data categories, and current detection rules are input into the tuning model to obtain updated detection rules. Simultaneously, a standard test set is acquired, and the updated detection rules are used to process the standard test set to obtain a first detection result. The updated detection rules are then used to process the standard test set to obtain a second test result. The first test result and the updated detection rules, as well as the second test result and the updated detection rules, are displayed on a screen. A tuning selection instruction is received, and the model parameters of the tuning model are adjusted according to the tuning selection instruction. The tuning process is repeated using the tuning model corresponding to the adjusted model parameters until the tuning selection instruction indicates that the update of the detection rules should be stopped. The updated detection rules are then saved to the current rule group. This method effectively addresses the shortcomings of cumbersome, costly, and poorly scalable optimization processes in detection rule tuning, significantly improving the convenience and responsiveness of dynamic rule tuning.

[0164] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0165] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0166] 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.

[0167] 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.

[0168] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An adaptive detection rule optimization method, characterized in that, The method includes: Receive tuning instructions, determine rule groups and corresponding current detection rules based on the tuning instructions, and receive tuning data and data categories corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between the text data and the image data. The data categories include positive example category, negative example category and suspected example category. The optimization data, the data category, and the current detection rule are input into the optimization model to obtain the updated detection rule. A standard test set is obtained, and the standard test set is processed by the updated detection rule to obtain a first test result. The standard test set is then processed by the detection rule to obtain a second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time. The first test result and the updated detection rule, the second test result and the detection rule are displayed on the display screen. An optimization selection instruction is received. Based on the optimization selection instruction, the model parameters of the optimization model are adjusted. The steps of inputting the optimization data, the data category and the current detection rule into the optimization model to obtain the updated detection rule are repeated through the optimization model with the adjusted model parameters until the optimization selection instruction indicates to stop updating the detection rule. The updated detection rule is then saved to the current rule group.

2. The method according to claim 1, characterized in that, The received optimization data and the data categories corresponding to each optimization data include: Receive image data or text data, and send annotation prompt information when the image data or text data is received, and receive prompt feedback information corresponding to the annotation prompt message. The annotation prompt information is used to prompt the user to upload the text data corresponding to the image data, or to upload the image data corresponding to the text data. The image data and the corresponding text data are defined as a set of tuning arrays, and the tuning data includes one or more sets of tuning arrays.

3. The method according to claim 1, characterized in that, The step of inputting the tuning data, the data category, and the current detection rule into the tuning model to obtain the updated detection rule includes: The optimization data is preprocessed, and the features of the preprocessed optimization data are extracted to obtain optimization data features, which include text data features and image data features. The text data features are input into the text branch of the optimization model for processing to obtain text processing data. The image data features are input into the image branch of the optimization model for processing to obtain image processing data. The rule data features corresponding to the current detection rule stored in the optimization model are retrieved. The weights corresponding to the text processing data, image processing data, and rule data features are dynamically adjusted by a gating attention mechanism, and a weighted average is performed on the text processing data, image processing data, and rule data features to obtain the updated detection rule.

4. The method according to claim 1, characterized in that, The step of adjusting the model parameters of the tuning model based on the tuning selection instruction includes: The optimization result for the current round is determined based on the optimization selection instruction, and the optimization result is either the first test result or the second test result; If the optimization result is the first test result, the first test result is determined as the reward signal for the current round, and the second test result is determined as the penalty signal for the current round. If the optimization result is the second test result, the second test result is determined as the reward signal for the current round, and the first test result is determined as the penalty signal for the current round. The learning rate of the tuning model is increased based on the reward signal, and the learning rate of the tuning model is decreased based on the penalty signal, so as to adjust the model parameters of the tuning model.

5. The method according to claim 1, characterized in that, The tuning instructions include adopting update instructions, continuing optimization instructions, and rolling back version instructions; after receiving the tuning selection instructions, it also includes: When the optimization selection instruction is the adoption update instruction, the updating of the detection rules and the model parameters of the optimization model shall be stopped; When the optimization selection instruction is the continue optimization instruction, the step of adjusting the model parameters of the optimization model based on the optimization selection instruction is executed; When the optimization selection instruction is the rollback version instruction, the detection rule corresponding to the previous iteration round is retrieved, and the detection rule corresponding to the previous iteration round is determined as the updated detection rule.

6. The method according to claim 1, characterized in that, The receiving of tuning instructions includes: Receive tuning instructions through the interactive interface on the display screen; After receiving the tuning instruction, the following is also included: The basic information of the current detection rule and the historical performance data of the current detection rule are displayed on the display screen of the terminal device. The basic information includes the rule name and rule description of the detection rule. The historical data to be detected within a preset time period prior to the current time, processed by the detection rules, and the corresponding historical detection results are displayed on the display screen.

7. An adaptive detection rule optimization device, characterized in that, The device includes: The receiving module is used to receive tuning instructions, determine rule groups and the current detection rules corresponding to the rule groups based on the tuning instructions, and receive tuning data and the data categories corresponding to each tuning data. The tuning data includes text data and image data, and there is a one-to-one correspondence between the text data and the image data. The data categories include positive example categories, negative example categories and suspected example categories. The testing module is used to input the optimization data, the data category and the current detection rule into the optimization model to obtain the updated detection rule, obtain a standard test set, process the standard test set through the updated detection rule to obtain a first test result, and process the standard test set through the detection rule to obtain a second test result. The standard test set includes historical sample data and optimization sample data received within a preset time period before the current time. The tuning module is used to display the first test result and the updated detection rule, the second test result and the detection rule on the display screen, receive a tuning selection instruction, adjust the model parameters of the tuning model based on the tuning selection instruction, and repeatedly execute the step of inputting the tuning data, the data category and the current detection rule into the tuning model to obtain the updated detection rule through the tuning model with the adjusted model parameters, until the tuning selection instruction indicates to stop updating the detection rule, and save the updated detection rule to the current rule group.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the adaptive detection rule optimization method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the adaptive detection rule optimization method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the adaptive detection rule optimization method according to any one of claims 1 to 6.

Images